Banking Technology & Allied Topics

— Automate governance and controls to ensure internal talent pool of data scientists and engineers. business-and-technology teams have ready However, most treat data as an operational function access to appropriate data sets, with the and leverage data-and-analytics talent primarily necessary controls for security and permission to generate and automate reports required by where needed. It is also important to ensure traditional business teams. A few leaders treat data that the appropriate data are available for management as a strategic function, and embed decisioning, at the right time and in the right data scientists/engineers within agile product form, to the various AA/ML models used and customer service teams, each focused on a by internal teams (from customer service to discrete journey or use case, such as small business product management) to support intelligent, lending, home financing, or digital wealth advisory highly personalized interactions with for the mass affluent. These organizations have customers. been recognized as leaders in creating superior experiences that give them a competitive edge, 2. Embed next-generation talent within traditional measured in customer satisfaction and value teams. Creating superior customer experiences creation. in the digital era requires a new set of skills and capabilities centered on design, data science, 3. Institute formal top-down mechanisms to and product management. An individual product support coordination across traditional product and manager, for example, may focus primarily on channel silos. While financial services institutions technical solutions, customer experiences, or take various measures to align working teams with maximizing business performance, but in an groups focused on serving a specific customer AI-first environment, all product managers will segment, these measures typically take a long time need a foundation in diverse areas, including to yield results (and often fail). The product and customer experience, advanced analytics and channel silos through which banks have traditionally machine learning, market analysis, business sought to address the needs of diverse market strategy, as well as leadership and capability segments can be very complex, and this complexity development.⁸ Design leaders require a similar makes it difficult to break out of the product-centric foundation as well as deep expertise in extracting mindset and assume a genuinely customer-centric user insights to guide business strategy and view throughout the organization. innovation.⁹ The data, analytics, and AI skills required to build an AI-bank are foreign to most In our experience, bottom-up efforts to organize traditional financial services institutions, and teams around customer segments often fall short organizations should craft a detailed strategy of expectations if they are not complemented by a for attracting them. This plan should define top-down approach consisting of cross-department which capabilities can and should be developed senior management teams. While these teams are in-house (to ensure competitive distinction) and empowered to act (that is, they have resources and which can be acquired through partnerships with budgets, along with autonomy in deciding how to technology specialists. deploy these to meet strategic goals), they also take an integrated view of various siloed efforts across Furthermore, our experience suggests that it’s the organization and prioritize a limited number of not enough to staff the teams with new talent. high-impact cross-cutting initiatives that require What really differentiates experience leaders central coordination (as opposed to spreading the is how they integrate new talent in traditional organization’s resources thin on several smaller team structures and unlock the full potential of initiatives). Finally, they develop and track progress these capabilities, in the context of business against a coordinated plan executed through the problems. Several organizations have built an traditional team structure. 8 Chandra Gnanasambandam, Martin Harrysson, Shivam Srivastava, and Yu Wu, “Product managers for the digital world,” May 2017, McKinsey.com. 9 Melissa Dalrymple, Sam Pickover, and Benedict Sheppard, “Are you asking enough from your design leaders,” February 2020, McKinsey.com. 27 Reimagining customer engagement for the AI bank of the future

4. Institutionalized capabilities to strike new and proof-of-concept trials, as well as partnerships at-scale with a heterogenous set of modern data-sharing and storage options non-financial services institutions. Partnerships compatible with the partner’s data-stack. are becoming increasingly critical for financial services players to extend their boundaries 5. Deep integration with the remaining beyond traditional channels, acquire more layers of the AI bank—that is, the AI-enabled customers, and create deeper engagement. Most decisioning layer and the core-tech and data institutions understand the importance of having layer. The journey to become an AI bank a clear strategic rationale (including a “win-win” entails transforming capabilities across all four value creation thesis for partners), and a strong layers of the capability stack: engagement, governance model to oversee the partnership. It AI-powered decisioning, core technology and is also important to establish teams responsible data infrastructure, and operating model. The both for setting up partnerships and for adapting layers should work in unison, and investment in the technology infrastructure to support the each layer should be made in tandem with the efficient and speedy launch of the partnership. others. Underinvesting in any layer will create a ripple effect that hinders the ability of the stack — Setting-up dedicated teams that are focused as a whole to deliver enterprise goals. on establishing partnerships. These teams constantly scan the market for potential As traditional banks observe the rapid partners and assess their relevance to the advancement of AI technologies and the institution’s growth strategy. They engage success of digital innovators in creating effectively with a broad range of non- compelling customer experiences, many bank partners—beginning with a review of recognize the need to reimagine how they differences in culture and technology—and engage their customers. By adopting an gauge the flexibility required to align with the AI-first approach in their vision and planning, partners’ ways of working (e.g., profile and innovative banks are building the capabilities seniority of people participating in discussions, that will enable them not just to deliver decision-making styles, responsiveness to intelligent services but also to design intuitive, requests, adherence to timelines) to enable highly personalized journeys spanning diverse faster, smoother, and more productive ecosystems, from banking to housing to collaboration. retail commerce, B2B services, and more. To realize this vision requires new talent, a robust — Making the technology infrastructure mechanism for managing partnerships, and a partnership-friendly hinges to a significant progressive transformation of the capability degree on API contracts identifying the stack. Throughout this expansive undertaking, functionalities that must be developed to meet leaders must stay attuned to customer the partner’s requirements. Another crucial perspectives and be clear about how the AI step is altering the technology infrastructure bank will create value for each customer. to facilitate fast integration with partner capabilities. This includes creating sand-box environments to enable rapid experimentation Renny Thomas is a senior partner, Shwaitang Singh is an associate partner, and Sailee Rane is a consultant, all in McKinsey’s Mumbai office. Malcom Gomes is a partner in the Bengaluru office, and Violet Chung is a partner in the Hong Kong office. The authors would like to thank Xiang Ji, Jinita Shroff, Amit Gupta, Vineet Rawat, and Yihong Wu for their contributions to this article. Reimagining customer engagement for the AI bank of the future 28

Global Banking & Securities AI-powered decision making for the bank of the future Banks are already strengthening customer relationships and lowering costs by using artificial intelligence to guide customer engagement. Success requires that capability stacks include the right decisioning elements. by Akshat Agarwal, Charu Singhal, and Renny Thomas March 2021 © Getty Images 29

The ongoing transition to digital channels creates — Lower operating costs. Banks can lower costs an opportunity for banks to serve more customers, by automating as fully as possible document expand market share, and increase revenue at lower processing, review, and decision making, cost. Crucially, banks that pursue this opportunity particularly in acquisition and servicing. also can access the bigger, richer data sets required to fuel advanced-analytics (AA) and machine- — Lower credit risk. To lower credit risks, banks learning (ML) decision engines. Deployed at scale, can adopt more sophisticated screening of these decision-making capabilities powered prospective customers and early detection of by artificial intelligence (AI) can give the bank a behaviors that signal higher risk of default and decisive competitive edge by generating significant fraud. incremental value for customers, partners, and the bank. Banks that aim to compete in global As banks think about how to design and build a and regional markets increasingly influenced highly flexible and fully automated decisioning by digital ecosystems will need a well-rounded layer of the AI-bank capability stack, they can AI-and-analytics capability stack comprising four benefit from organizing their efforts around four main layers: reimagined engagement, AI-powered interdependent elements: (1) leveraging AA/ML decision making, core technology and data models for automated, personalized decisions infrastructure, and leading-edge operating model. across the customer life cycle; (2) building and deploying AA/ML models at scale; (3) augmenting The layers of the AI-bank capability stack are AA/ML models with what we call “edge” capabilities¹ interdependent and must work in unison to deliver to reduce costs, streamline customer journeys, and value, as discussed in the first article. In our enhance the overall experience; and (4) building second article, we examined how AI-first banks an enterprise-wide digital-marketing engine to are reimagining customer engagement to provide translate insights generated in the decision-making superior experiences across diverse bank platforms layer into a set of coordinated messages delivered and partner ecosystems. Here, we focus on the AA/ through the bank’s engagement layer. ML decisioning capabilities required to understand and respond to customers’ fast-evolving needs with Automated, personalized decisions precision, speed, and efficiency. Banks that leverage across the customer life cycle machine-learning models to determine in (near) real time the best way to engage with each customer If financial institutions begin by prioritizing the use have potential to increase value in four ways: cases where AA/ML models can add the most value, they can automate more than 20 decisions in diverse — Stronger customer acquisition. Banks gain an customer journeys. Within the lending life cycle, for edge by creating superior customer experiences example, leading banks are relying increasingly with end-to-end automation and using advanced on AI and analytics capabilities to add value in five analytics to craft highly personalized messages main areas: customer acquisition, credit decisioning, at each step of the customer-acquisition journey. monitoring and collections, deepening relationships, and smart servicing (Exhibit 1, next page). — Higher customer lifetime value. Banks can increase the lifetime value of customers Customer acquisition by engaging with them continuously and The use of advanced analytics is crucial to the intelligently to strengthen each relationship design of journeys for new customers, who may across diverse products and services. follow a variety of paths to open a new card account, 1Edge capabilities refer to next-generation AI-powered technologies that can provide financial institutions an edge over the competition. Natural language processing (NLP), voice-script analysis, virtual agents, computer vision, facial recognition, blockchain, robotics, and behavioral analytics are some of the technologies that we classify as “edge capabilities.” These capabilities can be instrumental in improving customer experience and loyalty across multiple dimensions (engagement channels, intelligent advisory, faster processing), personalizing offers with highly accurate underwriting, and improving operational efficiency across the value chain (from customer servicing to monitoring, record management, and more). 30 AI-powered decision making for the bank of the future

Exhibit 1 Banks should prioritize using advanced analytics (AA) and machine learning (ML) in decisions across the customer life cycle. Customer life cycle Customer Credit decisioning Monitoring and Deepening Smart servicing acquisition Credit qualification collections relationships Hyperpersonalized Limit assessment Intelligent offers (eg, Servicing personas Pricing optimization Early-warning next product to buy) offers Fraud prevention signals Dynamic customer Churn reduction routing (channel, agent) Customer retargeting Probability of default/self-cure Channel propensity Real-time recom- Propensity-to-buy mendation engine scoring VAR-based customer Fatigue rule engine AI-enabled agent segmentation review and training Channel mapping Agent–customer mapping Monthly customer- Credit-approval Average days past Deposit/AUM Net promoter score, acquisition run rate turnaround time, % due, nonperforming attrition rate, cost of servicing products 6 per of applications assets approved customer Key metrics 1VAR is value at risk. AUM is assets under management. apply for a mortgage, or research new investment scoring. Based on real-time analysis of a customer’s opportunities. Some may head directly to the bank’s digital footprint, banks can display a landing page website, mobile app, branch kiosk, or ATM. Others tailored to their profile and preferences. may arrive indirectly through a partner’s website or by clicking on an ad. Many banks already use analytical These tools can also help banks tailor follow-up tools to understand each new customer’s path to the messages and offers for each customer. Replacing bank, so they get an accurate view of the customer’s much of the mass messaging that used to flow to context and direction of movement, which enables thousands or tens of thousands of customers in a them to deliver highly personalized offers directly subsegment, advanced analytics can help prioritize on the landing page. Following local regulations customers for continued engagement. The bank can governing the use and protection of customer data, select customers according to their responsiveness banks can understand individuals’ needs more to prior messaging—also known as their “propensity precisely by analyzing how customers enter the to buy”—and can identify the best channel for website (search, keywords, advertisements), their each type of message, according to the time of browsing history (cookies, site history), and social- day. And for the “last mile” of the customer journey, media data to form an initial profile of each customer, AI-first institutions are using advanced analytics including financial position and provisional credit to generate intelligent, highly relevant messages AI-powered decision making for the bank of the future 31

and provide smart servicing via assisted channels to usage data, and more. This decisioning process create a superior experience, which has been shown to is automated from end to end, so it can be contribute to higher rates of conversion.² completed nearly instantaneously, enabling the bank to predict the likelihood of default for Credit decisioning individuals in a vast and potentially profitable Setting themselves apart from traditional banks, segment of unbanked and underbanked whose customers may wait anywhere from a day to a consumers and SMEs. As banks build and refine week for credit approval, AI-first banks have designed their qualification model, they can proceed streamlined lending journeys, using extensive gradually, testing and improving the model—for automation and near-real-time analysis of customer example, by using auto-approvals for customers data to generate prompt credit decisions for retailers, up to a certain threshold with significantly lower small and medium-size enterprises (SMEs), and default risk and using manual verification to corporate clients. They do this by sifting through a review those estimated to have a higher default variety of structured and unstructured data collected risk and then gradually shifting more cases to from conventional sources (such as bank transaction automated decisioning. history, credit reports, and tax returns) and new ones (including location data, telecom usage data, utility — Limit assessment. Leading banks are also using bills, and more). Access to these nontraditional data AA/ML models to automate the process for sources depends on open banking and other data determining the maximum amount a customer sharing guidelines as well the availability of officially may borrow. These loan-approval systems, by approved APIs and data aggregators in the local leveraging optical character recognition (OCR) market. Further, while accessing and leveraging to extract data from conventional data sources personal data of customers, banks must secure data such as bank statements, tax returns, and and protect customer privacy in accordance with utilities invoices, can quickly assess a customer’s local regulations (e.g., the General Data Protection disposable income and capacity to make regular Regulation in the EU and the California Consumer loan payments. The proliferation of digital Privacy Act in the US). interactions also provides vast and diverse data sets to fuel complex machine-learning models. By using powerful AA/ML models to analyze these By building data sets that draw upon both broad and diverse data sets in near real time, banks conventional and new sources of data, banks can qualify new customers for credit services, can generate a highly accurate prediction of determine loan limits and pricing, and reduce the risk a customer’s capacity to pay. Just a few data of fraud. sources that may be available for analysis (with the customer’s permission) are emails, SMS, and — Credit qualification. Lenders seeking to determine e-commerce expenditures. if a customer qualifies for a particular type of loan have for many years used rule-based or — Pricing. Banks generally have offered highly logistic-regression models to analyze credit standardized rates on loans, with sales bureau reports. This approach, which relies on representatives and relationship managers a narrow set of criteria, fails to serve a large having some discretion to adjust rates within segment of consumers and SMEs lacking a formal certain thresholds. However, fierce competition credit history, so these potential customers turn on loan pricing, particularly for borrowers with a to nonbank sources of credit. In recent years, strong risk score, places banks using traditional however, leading banks and fintech lenders approaches at a considerable disadvantage have developed complex models for analyzing against AI-and-analytics leaders. Fortified with structured and unstructured data, examining highly accurate machine-learning models for risk hundreds of data points collected from social scoring and loan pricing, AI-first banks have been media, browsing history, telecommunications able to offer competitive rates while keeping their 2Erik Lindecrantz, Madeleine Tjon Pian Gi, and Stefano Zerbi, “Personalizing the customer experience: Driving differentiation in retail,” April 2020, McKinsey.com. 32 AI-powered decision making for the bank of the future

risk costs low. Some are also using their decisioning (e.g., fraud by sales agents), customer fraud, and capabilities to quantify the customer’s propensity payment fraud (including money laundering and to buy according to the customer’s use of different sanctions violations). Banks should continuously types of financial products. Some even leverage update their fraud detection and prevention natural-language processing (NLP) to analyze models, as we discuss later with regard to edge unstructured transcripts of interactions with sales capabilities. Ping An, for example, uses an image- and service representatives and, in some cases, analytics model to recognize 54 involuntary micro- collections personnel. By basing the offered rate expressions that occur before the brain has a on both creditworthiness and propensity to buy, the chance to control facial movements.³ bank can optimize the balance of total asset volume, risk, and interest income within a lending portfolio. AI-driven credit decisioning can build the business while lowering costs. Sharper identification of risky — Fraud management. As competition for credit customers enables banks to increase approval relationships becomes concentrated in digital rates without increasing credit risk. What is more, by channels, the automated processing of loan automating as much of the lending journey as possible, applications and use of AA/ML models to expedite banks can reduce the costs of support functions and credit approval and disbursement of funds not strengthen each customer’s experience with faster only positions the bank to acquire new customers loan approval and disbursement of funds, fewer and increase market share, but also opens new requests for documentation, and credit offers precisely opportunities for fraud. The costliest instances of tailored to meet customer needs. Exhibit 2 illustrates fraud typically fall into one of five categories: identity how AI-enabled decisioning capabilities underpin a theft, employee fraud, third-party or partner fraud customer’s onboarding journey. 3 “Chinese banks start scanning borrowers’ facial movements,” Financial Times, October 28, 2018, ft.com. Exhibit 2 The combination of AI and analytics enhances the onboarding journey for each new customer. Name: Joy Family: Married, no children Age: 32 years Profile attribute: Avid traveler Occupation: Working professional Propensity-to-buy model to identify AA-enabled real- Joy whom to retarget time credit underwriting, limit completes Joy’s Channel propensity Joy assessment, and online know-your- Hyperpersonalized cross-sell and landing page to identify right receives a call to customer form, upsell offers shows her per- outreach channel assist in journey; pricing provides details for Loan disbursed to Joy; curated sonalized offers: caller also employment verifi- catalog of offers personal loan for informs cation, and sets up ahead of Joy’s travel sent by travel, 5% off on Joy about custom an online pay- email travel insurance travel services Joy completes a ment mandate Joy receives a WhatsApp streamlined reminder for 3-click journey Analytics-backed personal loan Analytics-enabled to see the AI capabilities to hyperpersonalized for travel at zero customer–caller offer terms and conduct relevant mapping with fraud checks (eg, offers based on processing conditions charges customer specific cues facial recognition microsegment provided to caller with know-your- customer docs) AI-powered decision making for the bank of the future 33

Monitoring and collections variety of external data partnerships for location Once a bank has employed AA/ML models to data and transaction history can help the bank automate loan underwriting and pricing, it can also understand both the customer’s position and the deploy AI and advanced analytics to reduce the most effective approach, or contact strategy, for burden of nonperforming loans. Increasingly, banks averting default (Exhibit 3). are engaging with clients proactively to help them keep up with payments and work more closely Contact strategy. To determine an appropriate with clients who encounter difficulties. By drawing contact strategy for customers at risk of default, upon internal and external data sources to build a banks can segment accounts according to value 360-degree view of a customer’s financial position, at risk (VAR), which is the loan balance times banks can recognize early-warning signals that a the probability of default. This allows banks to borrower’s risk profile may have changed and that focus high-touch interactions on borrowers that the risk of default should be reassessed. account for the highest VAR; banks can then use low-cost channels like telephoning and texting for Beyond conventional data sources like repayment borrowers posing less risk. Banks have used this data and credit bureau reports, banks can digitize approach to reduce both the cost of collections and leverage other interaction data from campaigns, and the volume of loans to be resolved through field visits, and collection agents’ comments to restructuring, sale, or write-off.⁴ draw insights for collections strategy. Further, a 4Ignacio Crespo and Arvind Govindarajan, “The analytics-enabled collections model,” April 2018, McKinsey.com. Exhibit 3 Advanced analytics and machine learning can classify customers into microsegments for targeted interventions. Customer type True low-risk Absentminded Dialer-based True high-touch Unable to cure Targeted Use least Ignore or use Match agents to Focus on customers Offer debt- intervention experienced agents interactive voice customers; send live able to pay and at restructuring provided with set message (segment prompts to agents high risk of not settlements early for scripts will probably to modify scripts paying those truly self-cure) underwater Impact Onscreen prompts 10% of time saved, Matching and Added focus Significant increase guide agent–client allowing for conversation based reassignment of prompts can addresses higher in restructuring and on probability of agents to more breaking promises difficult customers increase sense of probability of default settlements and specific connection and rates in this segment increases chance of campaigns likelihood of paying collecting at least part of debt Source: Ignacio Crespo and Arvind Govindarajan, “The analytics-enabled collections model,” McKinsey on Payments, August 2018, McKinsey.com 34 AI-powered decision making for the bank of the future

Treatment strategy. If contact strategies through a prediction algorithm to estimate the ideal various channels are inadequate to help the product-per-customer (PPC) ratio for each customer resume timely payment, banks must user, based on individual needs. If analysis of pursue stronger measures, according to the a customer’s needs produces an anticipated customer’s ability and willingness to pay. Customers product usage ratio of eight but the customer with high willingness but limited ability to pay in the uses only two products, the relationship short term may require restructuring of the loan manager receives a prompt to reach out to the through partial-payment plans or loan extensions. customer and cross-sell or up-sell relevant In cases where the customer exhibits both low ecosystem products.⁵ willingness and limited ability to pay, banks should focus on early settlement and asset recovery. Servicing and engagement Advanced analytics, enabled by unstructured AI-powered decisioning can enable banks to internal data sources such as call transcripts create a smart, highly personalized servicing from collections contact centers and external experience based on customer microsegments, data sources such as spending behavior on other thereby enabling different channels to deliver digital channels, can improve the accuracy of superior service and a compelling experience determinations of ability and willingness to pay. with interactions that are fast, simple, and intuitive.⁶ Banks can support their relationship Deepening relationships managers with timely customer insights and Strong customer engagement is the foundation tailor-made offers for each customer. They can for maximizing customer value, and leaders also significantly improve agents’ productivity are using advanced analytics to identify less with streamlined preapproved products crafted engaged customers at risk of attrition and to craft to meet each customer’s distinct needs. Models messages for timely nudges. As with any customer that analyze voice and speech characteristics communication in a smart omnichannel service can match agents with customers based on environment, each personalized offer is delivered behavioral and psychological mapping. Similarly, through the right channel according to the time of transcript analysis can enable prediction of day. Rich internal data for existing customers can customer distress and suggest resolution to enable financial institutions to create a finely tuned the agent. outreach strategy for each individual customer, guided by risk considerations. Deployment of AA/ML models at scale Deeper relationships are predicated on a bank’s precise understanding of a customer’s unique Leveraging AI to automate decision making in needs and expectations. A bank can craft offers to near real time is a complex and costly endeavor. meet emerging needs and deliver them at the right If banks are to earn the required return on their time and through the right channel. By doing so, the technology investments, they must begin with a bank demonstrates that it understands customers’ strategy and road map to capture maximal scale current position and aspirations and can help them benefits in the design, building, and deployment get from the former to the latter. For example, by of AA/ML models. analyzing browsing history and spending patterns, a bank might recognize a consumer’s need for credit As banks embark on this journey, leaders must to finance an upcoming purchase of a household encourage all stakeholders to break out of siloed appliance. Analysis of internal data on product mindsets and think broadly about how models usage can also reveal areas where the bank can can be designed for uses in diverse contexts make its offering more relevant to a customer’s across the enterprise. AI-first organizations have current needs. Ping An, for example, has developed succeeded by organizing the effort around four 5“Ping An Bank: Change everything,” Asiamoney, September 26, 2019, asiamoney.com. 35 6Violet Chung, Malcolm Gomes, Sailee Rane, Shwaitang Singh, Renny Thomas, “Reimagining customer engagement for the AI bank of the future,” McKinsey.com, October 2020. AI-powered decision making for the bank of the future

main elements: First, they prioritize the analytics environment where cross-functional teams can use cases with the biggest impact on customer experiment with different approaches to achieving experience and the most value for the bank. the value-generating goals of a particular use case, Second, they ensure that the data architecture, moving from minimal viable product to scalable data pipelines, application programming interfaces solution in a matter of weeks. Building AA/ML models (APIs), and other essential components are available at scale and deploying them across the enterprise for building and deploying models at scale through depend on matching the right talent and skills with standardized, repeatable processes.⁷ Third, they each of the roles required for a successful analytics establish a semiautonomous lab for experimentation lab and factory (Exhibit 4).8 and prototype development and set up a factory for industrial-scale production of the solution. Fourth, The lab combines talent from business, analytics, they assemble the right mix of talent for agile, cross- technology, operations, and more. There are two functional teams and empower them to maximize main technical roles. One is the data scientist, value in close alignment with enterprise strategy. who is responsible for identifying the analytics techniques required to meet the business goal and Several leading banks have established semi- for programming advanced analytics algorithms. autonomous labs offering a test-and-learn The other is the data engineer, who scopes the data 7Tara Balakrishnan, Michael Chui, Bryce Hall, and Nicolaus Henke, “The state of AI in 2020,” November 2020, McKinsey.com. 8Nayur Khan, Brian McCarthy, and Adi Pradhan, “Executive’s guide to developing AI at scale,” October 2020, McKinsey.com. Exhibit 4 Diverse roles are necessary for building and deploying AA/ML models at scale. Lab environment Factory environment Product owner Data scientist Product owner ML engineer Leads the squad; Frames the First point of Optimizes ML typically a business problem contact for models for business owner and develops external stake- performance and who provides voice advanced analytics holders; defines scalability; deploys of the customer algorithms the solution criteria the models into production Data engineer Translator DevOps engineer Scopes the data Interfaces Develops CI/CD Infrastructure available; builds between business pipelines to architect data architecture and technical automate parts of Designs and data pipelines stakeholders the software- infrastructure deployment pipeline components for the analytics use case Full-stack Designer Delivery manager developer Focuses on Responsible for all Develops software interaction aspects of delivery components for between end of the analytics the back and front users and the solution to meet ends of AI solutions analytics solution squad goal output 1Continuous integration and continuous deployment. 36 AI-powered decision making for the bank of the future

available, identifies major sources of data to be Some edge technologies already afford banks consolidated for analytics, develops data pipelines to the opportunity to strengthen existing models simplify and automate data movement, and sets up with expanded data sets. For example, many data architecture for storage and layering. In addition, interactions with customers—via telephone, the role of translator is crucial to ensure consistent mobile app, website, or increasingly, in a communication and smooth collaboration between branch—begin with a conversational interface business leaders and analytics specialists. to establish the purpose of the interaction and collect the information required to resolve the On factory teams, one of the primary technical query or transfer it to an agent. A routing engine roles is the DevOps engineer, who is responsible can use voice and image analysis to understand for developing continuous integration (CI) and a customer’s current sentiment and match the continuous deployment (CD) pipelines for deploying customer with a suitable agent. The models software. In addition, the full-stack developer is underpinning virtual assistants and chatbots responsible for developing software components employ NLP and voice-script analysis to increase for other layers of the stack. The machine-learning their predictive accuracy as they churn through engineer prepares models for deployment at scale, vast unstructured data generated during and the infrastructure architect ensures that the customer-service and sales interactions. analytics solution is compatible with the architecture of the core tech and data layer of the capability stack. While each customer-service journey presents an opportunity to deepen the relationship with the The lab-and-factory setup requires flexible and help of next-product-to-buy recommendations, scalable technologies to handle the changing banks should constantly seek to improve their requirements of analytics engines. It is also important recommendation engines and messaging to give analytics teams access to the centralized campaigns. Feedback loops, for example, can data lake, and these teams must be able to draw help marketing teams and frontline officers upon raw data from diverse sources to generate data gauge the effectiveness of an offer by analyzing sets to be used in building models. The technology customers’ ongoing browsing and transaction supporting the solution must be modular to allow the activity within the bank’s digital ecosystem and transfer of developed solutions to factory production beyond (Exhibit 5, next page). using DevOps tools. Finally, it is crucial to embed performance management and risk controls within As edge capabilities become more powerful, models to avoid adverse impacts on operations. leaders are developing new, increasingly complex analytics solutions to create a superior Once the lab has developed a model, the factory experience and introduce distinctive innovations. takes over, running 24/7 to put the model into Use of computer vision and voice-to-script production and deploying it at scale in diverse use conversion can speed the completion of forms— cases across the enterprise. for instance, enabling a customer to respond orally to questions and upload documents Augmented AA/ML models with edge from which relevant data can be extracted capabilities automatically using optical character recognition (OCR). Facial and sentiment analysis during an The rapid improvement of AI-powered technologies in-person consultation or videoconference can spurs competition on speed, cost, experience, and support frontline representatives with messages intelligent propositions. To maintain its market and offers finely tuned to the customer’s needs leadership, an AI-first institution must develop and aspirations. models capable of meeting the processing requirements of edge capabilities, including natural- Several banks use voice recognition to verify language processing (NLP), computer vision, facial customer identity for certain low-value, high- recognition, and more. volume transactions. Some are using facial AI-powered decision making for the bank of the future 37

Exhibit 5 Edge capabilities enhance customer-service journeys. Digital self-cure Voice-recognition- AI-enabled Voice-analytics- AI-enabled Feedback loop via service-to-sales engagement channels for enabled IVR, customer profiling, and NLP-enabled channels engine customers (eg, frontline bots customer–agent customer- WhatsApp, mobile handling 30–50% matching sentiment analysis app, website) of queries Customer may If query is not Customer Contact-center Customers are Postcall feedback prompted for any and automated seek immediate resolved through connected to the agents supported specific offers follow-up occur via resolution of automated appropriate agent by live feedback preapproved for digital channels queries through channels, (chat or call) based and prompts to them digital self-service customer may on type of query sustain superior channels contact bank via and customer customer chat or request profiling experience call with live agent 1Interactive voice response. Natural-language-processing-enabled. recognition to authenticate customers’ identity as soon of customer behaviors, banks must go the last mile to as they enter a branch, approach an ATM, or open the ensure that these analytical insights have an impact banking app on a mobile device. As noted earlier, facial on customer behavior, such as purchasing a product, analysis is also useful in identifying making a loan payment, or exploring new service potential fraud. offers. In other words, an organization must establish a mechanism to “translate” analytical outcomes Leading banks are using blockchain to create smart into compelling messages to communicate to the contracts, secure trade documents and automate the customer at the right time, through the preferred release of funds upon delivery of goods, and establish channel—be it email, SMS, mobile app, website, shared utilities to reduce the burden of know-your- branch staff, or a relationship manager—according customer (KYC) and anti-money-laundering (AML) to the time of day. compliance for banks and customers. Edge capabilities deployed as part of an enterprise strategy to enhance This last mile from decisioning to messaging is the the AI bank’s value proposition have the potential not domain of the digital marketing engine. Seamlessly only to improve credit underwriting and fraud integrated with applications across the full AI-and- prevention but also to reduce the costs of document analytics capability stack with the help of APIs, handling and regulatory compliance. from data infrastructure to engagement channels, this engine supports the nearly instantaneous Enterprise-wide digital marketing engine processing of raw data to produce tailored messages communicated via engagement channels. Exhibit 6, While the automated decisions generated by AA/ML on the next page, illustrates the position of the digital models provide highly accurate, real-time predictions 38 AI-powered decision making for the bank of the future

Exhibit 6 The digital marketing engine requires full stack capabilities. Engagement layer Email SMS Mobile app Website Branch Contact ... center Decisioning layer Martech stack Design and activation Channel analytics/ Customer Data mgmt. Content Ad tech Campaign feedback loop relationship platform mgmt. server mgmt. management (DMP) platform platform Testing platform (CRM) AA/ML models: Decisioning and personalization engine Edge NLP Voice Virtual Computer Facial Blockchain Robotics Behavioral capabilities analysis agents vision recognition analytics Data lake layer Curated data lake Raw data lake Other use cases Customer 360 Personalization store Additional derived elements for personalization Comprehensive set of data sources: internal structured data (eg, applications, product holding, payment behavior), unstructured data (eg, call logs), CRM, external data (eg, telco, clickstream data), campaign performance data 1Advanced-analytics and machine-learning. Natural-language processing. marketing engine (or martech stack) within the other interventions are created, managed, decisioning layer of the AI-bank capability stack. and modified; (2) the ad tech server, which automates advertisements based on data The digital marketing engine comprises platforms analysis; and (3) the campaign management and applications fulfilling four main functions: data platform, which supports the creation and management, design and activation, measurement management of marketing campaigns, which and testing, and channel analytics. The data are conducted automatically according to the management platform, which forms part of the microsegmentation generated by the data core tech and data infrastructure layer of the management platform. AI-bank capability stack, supplies the data used to create and manage target customer segments. Just as the AI-and-analytics capability The design and activation function has three stack entails fundamental changes in the elements: (1) the content management platform, organization’s talent, culture, and ways of where messages, offers, advertisements, and working, the success of digital marketing AI-powered decision making for the bank of the future 39

capabilities depends on an agile operating As an example, Commonwealth Bank of Australia model. This model consists of autonomous cross- (CBA) leverages its mobile app to test messages functional teams (or pods) drawing upon the and learn within hours what works and what must talent of different parts of the enterprise, such be changed. This cadence enables rapid scaling of as business units, marketing, analytics, channels, campaigns to similar customer segments.9 operations, and technology. Each pod should also include representatives from partner organizations In measurement of campaigns’ impact, scientific crucial to the digital marketing effort—for example, rigor is crucial. To allow for precise measurement user-interface and user-experience designers, of the incremental value of the campaign, each who lay out the campaign’s look and feel and its target segment should include a control group of flow, and copywriters, who finalize the language customers excluded from the campaign. The tools of any intervention. The members of each pod and capabilities for evaluating the effectiveness of collaborate on developing, managing, and improving customer-engagement campaigns help employees engagement campaigns, and each member is across the organization understand how they can accountable for campaigns’ impact according to enhance their impact on individual customers and clearly defined key performance indicators (KPIs). add value to an AI-oriented culture. To achieve the desired outcome, an AI-first bank The rapid improvement of AI-powered technologies launching daily personalized communications to spurs competition on speed, cost, experience, and millions of customers must build tools for continual intelligent propositions. To remain competitive, testing and learning. The measurement and testing banks must engage customers with highly platform flags potential aspects of content or personalized and timely content to build loyalty. distribution to improve, thereby enabling teams to Personalized offers with tailored communication evaluate in real time the effectiveness of campaigns. delivered at the right time through the customer’s preferred channel can help banks maximize the Another source of continual feedback is channel lifetime value of each customer relationship and analytics, which includes tools and dashboards reinforce the organization’s market leadership. for real-time tracking of engagement across each To achieve these benefits, banks must build target segment. Every day, each pod leverages the AI-powered decisioning capabilities fueled by a rich channel analytics and measurement and testing mixture of internal and external data and augmented platforms to closely track various indicators, by edge technologies. The core technology and including delivery rates, email open rates, click- data infrastructure required to collect and curate through rates by channel for customers seeking increasingly diverse and voluminous data sets is the more information (the first call to action), conversion topic of the next article in our series on the AI-bank rates, and more. These diagnostics help members capability stack. of the pod experiment with potential enhancements to messages, advertisements, and campaign design. Akshat Agarwal is an associate partner in McKinsey’s Bangalore office. Charu Singhal a consultant and Renny Thomas is a senior partner, both in the Mumbai office. 9Paul McIntyre, “CommBank’s analytics chief on how its AI-powered ‘Customer Engagement Engine’ is changing everything,” Mi3, September 21, 2020, mi-3.com.au. 40 AI-powered decision making for the bank of the future

Global Banking & Securities Beyond digital transformations: Modernizing core technology for the AI bank of the future For artificial intelligence to deliver value across the organization, banks need core technology that is scalable, resilient, and adaptable. Building that requires changes in six key areas. by Sven Blumberg, Rich Isenberg, Dave Kerr, Milan Mitra, and Renny Thomas April 2021 © Getty Images 41

An artificial-intelligence (AI) bank leapfrogs An AI-first model places demands on the competition by organizing talent, technology, a bank’s core technology and ways of working around an AI-first vision for empowering customers with intelligent value Across industries, many organizations have propositions delivered through compelling struggled to keep pace with the demand for journeys and experiences. Making this vision digitization, especially as consumers accelerated a reality requires capabilities in four areas: their adoption of digital channels for daily an engagement layer, decisioning layer, core transactions during the COVID-19 crisis.¹ Even technology layer, and platform operating model. before that, however, the financial-services industry has historically had mixed success We discussed the the first two areas in the in technology. Institutions that were early previous articles. The capabilities of the adopters and innovators in technology have reimagined engagement layer enable the AI bank built up a complex landscape of technical assets to deliver highly personalized seamless journeys over decades and accumulated significant across bank channels and within partner technical debt. Some institutions have tackled ecosystems. The capabilities of the AI-powered this challenge; many are behind the curve. decisioning layer transform customer insights Meanwhile, alongside the incumbents, an into messages and offers tailored to address a extremely active fintech industry has been customer’s unique needs. The current article constantly innovating and raising the bar. identifies capabilities needed in the third area, the core technology and data infrastructure of Financial institutions that have shifted from the modern capability stack. being intensive consumers of technology to making AI and analytics a core capability are Deploying AI capabilities across the organization finding it easier to shift into the real-time and requires a scalable, resilient, and adaptable consumer-centric ecosystem. As AI technologies set of core-technology components. When play an increasingly central role in creating value implemented successfully, this foundational for banks and their customers, financial-services layer can enable a bank to accelerate technology organizations need to reinvent themselves innovations, improve the quality and reliability as technology-forward institutions, so they of operations, reduce operating costs, and can deliver customized products and highly strengthen customer engagement. personalized services at scale in near real time. We begin by summarizing the primary demands At many institutions, standard practices now banking leaders should consider as they plan include omnichannel engagement, the use of an enterprise-wide initiative to modernize APIs to support increased real-time information core technology, data management, and the exchange across systems, and the use of big underlying infrastructure. Next, we examine data analytics to improve credit underwriting, the key transformations required to modernize evaluate product usage, and prioritize the core technology and data infrastructure. opportunities for deepening relationships. As We conclude by sharing 12 actions technology financial-services organizations continue leaders should consider taking to ensure the to mature, the increasing demands on the transformation creates value for customers and technology infrastructure to support more the bank. complex use cases involving analytics and real- 1Tamara Charm, Anne Grimmelt, Hyunjin Kim, Kelsey Robinson, Nancy Lu, Mayank, Mianne Ortega, Yvonne Staack, and Naomi Yamakawa, “Consumer sentiment and behavior continue to reflect the uncertainty of the COVID-19 crisis,” October 2020, McKinsey.com. 42 Beyond digital transformations: Modernizing core technology for the AI bank of the future

time insights are pushing firms to reexamine demand and capacity to meet strategic and near- their overall technology function. Once they have term priorities, and a well-defined mechanism to committed to modernizing the core technology coordinate “change the bank” and “run the bank” and data infrastructure underpinning the initiatives according to their potential to engagement and decision-making layers of the generate value. capability stack, banks should organize their transformation around six crucial demands: Faster time to market requires efficient and technology strategy, superior experiences, repeatable development and testing practices scalable data and analytics platforms, scalable coupled with robust platforms and productivity- hybrid infrastructure, configurable product measurement tools. Aligning demand and capacity processors, and cybersecurity strategy (Exhibit 1). according to strategic priorities works on two levels. On one level, banks need to ensure that Robust strategy for building technology execution, infrastructure, and support capacity are capabilities optimized to ensure constant operation of all use Before embarking on a fundamental cases and journeys. On the other, with constant transformation of core technology and data uptime assured, work should be organized and infrastructure, financial-services organizations scheduled to expedite projects having the greatest should craft a detailed strategy for building impact on value. Finally, financial institutions an AI-first value proposition. They should also should establish clear mechanisms for setting develop a road map for the transformation, priorities and ensuring that each use case is focusing on three dimensions of value creation: designed and built to generate a return exceeding faster time to market with efficient governance capital investments and operating costs. and productivity tracking, clear alignment of Exhibit 1 The AI-bank transformation places several crucial demands on core technology and data infrastructure. Robust strategy for building Superior omnichannel journeys Modern, scalable platform for technology capabilities and customer experiences data and analytics Scalable hybrid infrastructure Highly configurable and Secure and robust strategy for the cloud scalable core product processors perimeter for access Beyond digital transformations: Modernizing core technology for the AI bank of the future 43

Superior omnichannel journeys and customer Modern, scalable platform for data and experiences analytics Building journeys that excite customers with Delivering highly personalized offers in near their speed, intuitiveness, efficiency, and impact real time requires AI-powered decision-making typically involves various applications spanning capabilities underpinned by robust data multiple bank and nonbank systems, all linked assets. What is more, the at-scale development together by a series of APIs and integrations. of machine-learning (ML) models that are This complex information exchange enables the context aware in real time requires automated organization to ingest valuable data from diverse DevSecOps² and machine-learning ops (MLOps) sources to produce highly personalized messages tools to enable secure and compliant continuous and offers that speak directly to the customer integration (CI) and continuous deployment (CD). in near real time. In addition to a standardized This entails complex orchestration across source approach to managing APIs, banks should develop systems, data platforms, and data sciences a clear mechanism to integrate across channels, to enable lab experimentation and factory core systems, and external interfaces while production. This is particularly complex in a highly managing changes across multiple dependent regulated environment where the involvement of systems. They should bear in mind, for example, security, audit, risk, and other functions is crucial that introducing a change in an existing digital in many stages of the process. channel could potentially entail changes not only across the front end but also across multiple The incorporation of feedback loops with channel interfacing systems, core product processors, and systems enables models to evaluate the output analytics layers. performance and make automated adjustments to increase the effectiveness of personalized A focus on journeys and user experience also messages, so the organization can generate benefits back-office and operations teams. New personalized offers nearly instantaneously. For products are increasingly automated at the back example, in the case of location-based offers end, freeing staff to focus on genuinely exceptional for adjacent products, an organization must be scenarios and differentiating activities, rather than able to overlay in real time customer location repetitive low-value activities. and preferences (as reflected in previous transactions) with predefined offers from nearby Finally, to ensure maximum value, use cases and participating merchants. capabilities should be designed as “enterprise products” to be reused in other areas. For example, Scalable hybrid infrastructure utilizing the cloud the deployment of microservices handling discrete With the continued expansion of customer tasks like document collection and ID verification engagement across bank and nonbank platforms, can ensure consistency in the way things are financial institutions need to create hyperscalable done across the organization. APIs should also infrastructure to process high-volume be documented and catalogued for reuse. APIs transactions in milliseconds. This capability that are domain- or product-centric (for example, is made possible, in part, by infrastructure enabling the retrieval of customer details from a as code, automated server provisioning, and single customer store) have higher reusability and robust automated configuration management take an enterprise-level view of the capability, as processes, which together solve the problem of compared with a journey-centric API design—for “snowflake” configurations resulting from organic example, one where an API supports retrieval of and complex linkages and changes that have customer details for a specific mobile journey. accumulated over time. 2DevSecOps tools support the integration of “development, security, infrastructure, and operations at every stage in the product’s life cycle, from planning and design to ongoing use and support.” See Santiago Comella-Dorda, James Kaplan, Ling Lau, and Nick McNamara, “Agile, reliable, secure, compliant IT: Fulfilling the promise of DevSecOps,” May 2020, McKinsey.com. 44 Beyond digital transformations: Modernizing core technology for the AI bank of the future

Hosting these environments on a distributed- as protection against vulnerabilities within network cloud environment allows a balance applications, operating systems, hardware, between paid-up-front baseline storage and and networks. Financial institutions should computing capacity, on the one hand, and, on the also implement appropriate measures to other, elastic on-demand surge capacity without secure the perimeter and control access disruptions to service. Self-monitoring and to various systems and applications within preventive maintenance also are automated, and the organization’s infrastructure footprint, disaster recovery and resiliency measures run in including private and public cloud servers the background to ensure constant uptime even and on-premises data centers. For example, if incidents evade automated self-repair and transferring workloads from traditional require manual intervention. As a result, the risk on-premises infrastructure to public cloud of disruption to critical operations is minimized, requires careful measures to protect customer and customer-facing applications run with high data, along with a robust strategy for detecting availability and responsiveness. The combination and remediating potential threats and of on-premises and cloud-based infrastructure vulnerabilities. is increasingly relevant in high-volume and high- frequency areas such as payments processing, The “classical” approaches of securing the core banking platforms, and customer perimeter should be coupled with more modern onboarding systems. Making workloads “cloud approaches to limit the impact of intrusions or native” and portable allows the work to be moved reduce the “blast radius.” Again, AI has a part to the most appropriate platform. to play here, given the advent of increasingly sophisticated network intrusion detection, Highly configurable and scalable core product anomaly detection, and even forensics during processors postmortems of security incidents. To sustain a leading-edge value proposition founded upon AI and ML capabilities, banks Start the transformation by must continually evaluate their core products prioritizing key changes and identify opportunities for innovations and customizations. Combined with deep To meet these demands, financial institutions understanding of customer needs, enabled will need to transition from a legacy by advanced analytics, an organization can architecture and operating model to an anticipate emerging customer requests and automation and cloud-first strategy. Building design distinctive products accordingly. The the core technology and data capabilities upon need for real-time reconciliation and round- a highly automated, hybrid-cloud infrastructure the-clock transaction processing also emerges can enable the AI bank to scale rapidly as a key competitive advantage for financial and efficiently as it gains competitive and institutions. For example, with the advent differentiating capabilities. of next-generation core banking platforms, organizations can now develop products that are The AI-bank capability stack combines core built for scale and can be readily configured to systems and AI-and-analytics capabilities in meet specific customer expectations.³ a unified architecture designed for maximal automation, security, and scalability. Getting Secure and robust perimeter for access to this target state requires a series of complex It is crucial to ensure that the organization initiatives to transform the organization’s core maintains an appropriate cybersecurity posture technology and data infrastructure. These across the entire technology infrastructure initiatives focus on several key areas: tech- 3Xavier Lhuer, Phil Tuddenham, Sandhosh Kumar, and Brian Ledbetter, “Next-generation core banking platforms: A golden ticket?” August 2019, McKinsey.com. Beyond digital transformations: Modernizing core technology for the AI bank of the future 45

forward strategy, modern API and streaming time-right releases. Organizations should also architecture, core processors and systems, data adopt enterprise agile practices for high-velocity management, intelligent infrastructure, and engineering teams, with integrated cross-functional cybersecurity and control tower (Exhibit 2). teams of business, technology, and functional experts, and external partners using modern Tech-forward strategy approaches to software development, testing, Banks should begin this far-reaching initiative by release, and support cycles. In addition, efficient translating the AI-first vision into an enterprise management of the full stack requires governance strategy that merges technology with business, of the technology function through a standardized funding investments in innovation with the returns set of metrics, along with ongoing tracking of on incremental changes in technology.⁴ Business uptime and health for each component of the stack. and technology collaborate as co-owners in designing and managing operating models Modern API and streaming architecture and outcomes. This “tech-forward” mindset Next, banks should integrate internal and external thrives in interdisciplinary teams focused on systems to support seamless customer journeys innovation and led by skilled engineering talent across internal platforms, partner ecosystems, leveraging modern tools and practices for first- and numerous external interfaces. This requires 4 The technology transformation described in this and other articles in our series on the AI bank of the future aligns broadly with our colleagues’ discussion of the tech-forward approach, which applies across industries. See Anusha Dhasarathy, Isha Gill, Naufal Khan, Sriram Sekar, and Steve Van Kuiken, “How to become tech-forward: A technology-transformation approach that works,” November 2020, McKinsey.com. Exhibit 2 Building a modern core technology and data infrastructure entails changes in several key areas. Channels and digital-journey integrations Tech-forward Modern APIs and Data management strategy streaming architecture for the AI world Core processors and systems Intelligent infrastructure Cybersecurity and control tower 46 Beyond digital transformations: Modernizing core technology for the AI bank of the future

a robust, scalable, and standardized approach models of the decision-making layer. The to building and hosting integrations and APIs. analytical insights generated by these models The APIs, in turn, should be rigorously tested for are deployed through martech tools to craft the performance and developed using agile release intelligent offers and smart experiences that set principles. When a well-defined stock of APIs-as- an AI bank apart from traditional incumbents. In products are orchestrating flows across systems, order to support superior omnichannel customer product innovations can advance from concept journeys and seamless integration with partner to production and deployment of minimum viable ecosystems, the data platform must be capable product within 30 to 60 days. of ingesting, analyzing, and deploying vast amounts of data in near real time. To complement a robust API strategy, technology leaders should also consider establishing a The data platform should also provide scalable high-speed data-streaming channel to enable workbenches with AI and data-science standardized asynchronous data transfer across capabilities to lab and factory teams. These the enterprise in real time. workbenches enable teams to access relevant data sets as they develop models and deploy Core processors and systems insights in product iterations. The infrastructure With the right architecture in place, banks can should also support the development of ML shift away from traditional, complex, and tightly models through automated and repeatable intertwined core systems to lightweight and processes. highly configurable core product processors and workflows. These processors are also If an organization allows interdisciplinary teams complemented by “microservices,” or discrete across the enterprise to search and extract applications (such as for payments, card data held on the platform, these teams can accounts, or loans) that “externalize” the logic optimize their data consumption according to within traditional core platforms. customer needs and market opportunities. It is essential to enable data-science teams with The transition to lightweight core processors and appropriate tooling and access to scalable systems hosted on scalable, modular, and lean computing power so that they may experiment platforms exposed as APIs supports, for example, and innovate. Underpinning these actions, real-time reconciliation and allows changes to be appropriate technical documentation and made in live systems with zero downtime. Use of cataloging of assets (for example, APIs, ML modern cloud-based infrastructure to host such models, data dictionary, DevOps and MLOps platforms also makes it easier to scale up. tools) ensure proper governance and access If successfully implemented, a lightweight control. By creating ML models and scorecards processor platform can enable an organization through a well-defined lab-factory model, to advance from new-product concept to launch AI-first organizations empower employees to in two to three months. This is a significant leverage self-serve, real-time data and analytics advantage against organizations constrained infrastructure to guide value-based planning by legacy technology, where launching a new and support daily decision making. product or customizing an existing product can take six months or more. Assembly of new Intelligent infrastructure off-the-shelf product stacks can also enable Banks then should ensure they have an innovative new customer propositions, such effective strategy to modernize infrastructure. as an end-to-end lending journey on a modern For this, they should consider the adoption stack using these principles. of public cloud to complement the traditional infrastructure in situations where workloads Data management for the AI world require resiliency, scale, and use of hosted It is crucial to establish a modern data and or managed offerings (such as hosted analytics platform to fuel the real-time ML databases). Public cloud enables velocity Beyond digital transformations: Modernizing core technology for the AI bank of the future 47

through higher levels of automation, templates, to maintain. Many organizations have spent billions and reduction of operational risk. When setting of dollars on multiyear technology initiatives up such environments, banks must build upon within silos, only to find that they fail to generate the foundational elements of infrastructure the scale benefits required to justify investments. management, including observability, resiliency, Leaders should heed these lessons, adopt a holistic and high availability, as well as a robust perspective, and map priorities according to the configuration strategy. A well-tuned, scalable, end-to-end impact that each step in the technology and load-balanced stack can support response transformation has on the value of the enterprise. times of less than a second while scaling horizontally to cater to variations in transaction If an organization meets the strategic demands volume. outlined at the top of this article, the implementation of modern core technology and data infrastructure Cybersecurity and control tower can yield significant value in the form of faster Finally, institutions should address cybersecurity delivery of changes and improvements, increased and control. This includes setting up a centralized cost efficiency, higher quality of assets, and control tower to monitor data, systems, and stronger customer outcomes. For example, a sound networks across the infrastructure. The scope DevOps and release-management strategy can of responsibility includes ensuring boundary contribute to a 25 to 30 percent increase in capacity security and identifying and rectifying threats creation, a reduction in time to market of 50 to 75 and intrusions. Also crucial is to establish a percent, and more than a 50 percent reduction well-defined set of compliance measures for in failure rates.⁵ In turn, development efforts can security testing and vulnerability scanning improve schedule adherence by 1.5 times and before deploying assets on live systems. These reduce customer defects by 20 to 30 percent measures reduce the risk posed by potential through process automation and agile ways of threat scenarios. working,⁶ and leading organizations have improved issue-resolution time and planning time by between Technology leaders should prioritize 30 and 50 percent.⁷ There are indirect benefits as interconnected capabilities well: by empowering employees with a clear mission, autonomy, and strong focus on customers, agile Given the broad scope of components to be organizations have been able to increase employee transformed, organizations should bear in engagement by 20 to 30 percent, as reflected both mind that optimal outcomes are much likelier in willingness to recommend their workplaces and in when they first establish a holistic strategy for employee-satisfaction surveys.⁸ technology transformation. Unfortunately, not all have found the resources to embrace fully the Technology transformations are fraught with risk, potential offered by the rapid advancement of including delays and cost overruns, and only those AI technologies and the steady rise in customer organizations whose leaders are prepared to expectations. Some financial institutions, commit the energy and capital necessary to carry despite seeing the imperative to change, have through with the comprehensive effort should maintained and modernized their legacy embark on the journey. Ultimately, this is a decision platforms. Various business lines have set up not just to survive, but to thrive, and it requires a organically built platforms upon this foundation, change in mindset. Specifically, traditional financial making it costlier and more and more complex institutions will need to break out of their legacy 5Thomas Delaet and Ling Lau, “DevOps: The key to IT infrastructure agility,” March 2017, McKinsey.com. 6Matt Brown, Ankur Dikshit, Martin Harrysosn, Shivam Srivastava, and Kunal Thanki, “A new management science for technology product delivery,” February 2020, McKinsey.com. 7Wouter Aghina, Christopher Handscomb, Jesper Ludolph, Daniel Rona, and Dave West, “Enterprise agility: Buzz or business impact?” March 2020, McKinsey.com. 8“Enterprise agility,” March 2020. 48 Beyond digital transformations: Modernizing core technology for the AI bank of the future

technology architecture and explore AI-and- enable repeatable execution and development analytics opportunities. Should they undertake of capabilities within technology teams and to the challenge and begin thinking about how promote standardization to speed up execution. best to chart their course to becoming an AI For example, a core system factory consisting bank, their leaders may consider 12 key insights of teams, predefined operating procedures, gleaned from the experience of financial- and systems to manage, prioritize, and execute services leaders that are in the process of changes across business units can expedite carrying out such transformations (Exhibit 3): deployment of new solutions significantly. 1. Consider the factory model to build at scale. 2. Consider insourcing differentiating capabilities. Leverage a factory approach in fast-evolving Based on the eventual outcomes desired, build and critical areas of the transformation to certain differentiating capabilities in-house, Exhibit 3 Leaders should consider 12 key insights as they embark on the technology- ttrraannssffoorrmmaattiioonn jjoouurrnneeyy.. Tech-forward strategy Consider the factory model to build at scale Consider insourcing differentiating capabilities Modern APIs and Maintain rigorous documentation on integrations streaming architecture Identify an anchor stack but experiment with others Core processors and systems Maintain automation-first and fast-release posture Consider a modern core for high-velocity areas Data management for Adopt a value-centric approach to building data platforms the AI world Set up a lab and factory for analytics Intelligent infrastructure Define the enterprise cloud strategy Establish end-to-end visibility across the stack Cybersecurity and control tower Identify the right perimeter design for the cloud Ensure data security on the cloud Beyond digital transformations: Modernizing core technology for the AI bank of the future 49

with robust engineering support, perhaps technology. We have observed that organizations starting with APIs, infrastructure, or the data that budget the anticipated return of change and analytics platform. efforts are able to prioritize use cases that are functionally simple, fit the road map for building 3. Maintain rigorous documentation on the platform in iterations, and realize economic integrations. Remember that the development value along the way. of engagement systems and comprehensive changes in core-technology require significant 8. Set up a lab and factory for analytics. Establish adjustments to integrations, and substandard a lab to experiment with tools and platforms for documentation of the specifications for these efficient development in test-and-learn cycles. integrations often slows the broader initiative Also, build a central factory for producing and to transform the bank. deploying analytics use cases at scale on an individual stack. 4. Identify an anchor stack but experiment with others. Emphasize the importance of 9. Define the enterprise cloud strategy. Create a standardization for engineering-centric common strategy across stakeholders to enable development at scale, and build on a single a structured and systematic migration to the stack to support faster change. At the same cloud. Cloud adoption poses multiple firsts in time, continue experimenting with other stacks the enterprise in terms of security perimeters, and stack components for smaller builds in change management, and cloud-migration and order to adopt alternative or newer approaches disposition strategy. where the incremental benefits are clearly defined. 10. Establish end-to-end visibility across the technology and infrastructure stack. Recognizing 5. Maintain an automation-first and fast- that at-scale digital transformations impose release posture. Adopt an automation-first limitations on volume and scale, implement robust and frequent-deployments posture on fast- automated tools to observe stack performance evolving applications and stacks. While initial and to diagnose and resolve issues. hiccups are not uncommon, release rails should be hardened over time to speed up time 11. Identify the right perimeter design for the cloud. to market. Well-defined release management To safeguard against potential malicious attacks and deployments are key to execution velocity. on cloud-based public-facing applications, Standardizing through DevSecOps typically design an appropriate network perimeter that unlocks productivity gains of as much as 20 to optimizes the potential attack radius. 30 percent. 12. Ensure data security on the cloud. Design 6. Consider a modern core for high-velocity robust data-categorization and data-security areas. Consider modern and lightweight safeguards to avoid critical customer-data core systems built on scalable and hybrid combinations and comply with national data- infrastructure to enable an efficient rollout of protection and data-residency laws. new capabilities while enabling a modular build of financial products. 7. Adopt a value-centric approach to building If banks are to thrive in a world where customer data platforms. Take advantage of the fact that expectations are increasingly shaped by the data and analytics platforms evolve over time, AI-and-analytics capabilities of technology leaders, and do not allow teams to be overwhelmed they must rebuild their core technology and data by the rapid shift of tooling and available infrastructure to support AI-powered decision 50 Beyond digital transformations: Modernizing core technology for the AI bank of the future

making and reimagined customer engagement. future requires an agile culture and platform- These are the three “technology layers” of the oriented operating model that respond promptly AI-bank capability stack. The full stack also to emerging opportunities and deliver innovative includes a leading-edge operating model to solutions rapidly at scale. The next article in ensure that all layers work together in unison to this series examines the crucial elements of the deliver intelligent propositions through smart platform operating model. servicing and experiences. The AI bank of the Sven Blumberg is a senior partner in McKinsey’s Dusseldorf office, Rich Isenberg is a partner in the Atlanta office, Dave Kerr is a senior expert in the Singapore office, Milan Mitra is an associate partner in the Bengaluru office, and Renny Thomas is a senior partner in the Mumbai office. The authors would like to thank Brant Carson, Kayvaun Rowshankish, Yihong Wu, and Himanshu Satija for their contributions to this article. Beyond digital transformations: Modernizing core technology for the AI bank of the future 51

Global Banking & Securities Platform operating model for the AI bank of the future Technology alone cannot define a successful AI bank; the AI bank of the future also needs an operating model that brings together the right talent, culture, and organizational design. by Brant Carson, Abhishek Chakravarty, Kristy Koh, and Renny Thomas May 2021 © Getty Images 52

As we noted at the beginning of this series on the and often several times a day through wearable AI bank of the future, disruptive AI technologies can devices. In short, banks and their customers now dramatically improve banks’ performance in four key have an interconnected, always-on relationship. areas: higher profits, at-scale personalization, smart omnichannel experiences, and rapid innovation Circumstances within the bank are changing as cycles. The stakes could not be higher, and success well—albeit at a slower pace, due largely to the requires a holistic transformation spanning all layers complexity of legacy technology and operating of the organization’s capability stack. models coupled with the steadily rising cost of maintaining and upgrading IT infrastructure. Siloed Our previous articles have focused on the capability structures also hamper organizations’ ability stack’s technology layers: reimagined engagement, to transform themselves. Decision making at AI-powered decision making, and modern core traditional banks is typically slow and cumbersome, technology and data infrastructure. Leveraging and ineffective prioritization (done at too high a these capabilities to create value requires an level without understanding underlying resource operating model combining structure, talent, culture, contentions) results in frequent project delays and ways of working to synchronize all layers of the and cost overruns. Insufficient domain expertise stack. Synchronizing these layers is not easy. Any and blurred accountability—particularly between organization undertaking an AI-bank transformation business units and technology teams—too often must determine how to structure the organization cause new solutions to fall short of customer so that its people interact and leverage tools and expectations. What is more, multiple systems capabilities to deliver value for each customer at perform similar functions, and the increasing scale. In this article, we take a closer look at the complexity of IT architecture with a proliferation of need for a platform operating model, the categories applications weakens system resilience and stability and scope of operating models, and the building and increases risk when changes are made. blocks of effective models. The widening divide between fast-evolving The heart of an AI bank is always-on customer expectations and inertia within the bank customer interaction reinforces silos and weakens the bank’s ability to respond to the demands of the new machine age. The need to change a bank’s operating model The challenge for leaders is to shift the organization arises from a combination of external and internal from this siloed structure to a radically flattened circumstances. Externally, as consumers and network of platforms. businesses increasingly rely on AI technologies in daily life, banks are shifting the foundation of their Platforms focus on delivering business business models from products to experiences. solutions In other words, as many traditional banking products become embedded—or even “invisible”— Today, banks that recognize the value of AI and within beyond-the-bank journeys, experiences technology enabling better customer and business become the more salient element of a customer’s experiences are moving steadily toward a platform relationship with the bank. This shift involves a rapid operating model, leveling command-and-control increase in the number of customer interactions, structures to speed decision making and bring and at the same time, the revenue associated with people together in teams relentlessly focused on each interaction is declining. This is a fundamental delivering solutions that customers value. In this change: just a few years ago, customers conducted agile approach, each platform can be thought of business with the bank by visiting a branch once or as a collection of software and hardware assets, twice a month; more recently, they would conduct funding, and talent that together provide a specific transactions several times each week through the capability. While some platforms, such as those bank website; now many customers interact with for retail mortgages, deliver business-technology their bank daily through their mobile banking app, solutions to serve internal or external clients, others 53 Platform operating model for the AI bank of the future

enable other platforms with shared services and — Organization and governance. Organization support functions (for example, payments and core of business-facing platforms (e.g., retail banking). Each platform is largely self-contained in mortgages) should be based on a “two in a producing business and technology outcomes and box” engagement model, meaning business autonomous in prioritizing its work to meet strategic and technology leaders own joint performance goals within clearly defined guardrails, such as metrics that track both commercial and common standards, finance, and risk control. technological outcomes. Each platform manages its business and technological Platform elements priorities through a shared backlog of work and As banks think about setting up a platform operating delivers through persistent cross-functional model, they should bear in mind that each platform agile teams, each of which builds its platform comprises three main elements. When structured over time and focuses not only on one project, correctly, these elements will help a platform team but continually improves the platform. set its North Star and carry out its mission in a way that creates value for customers and the enterprise. — Technology. Each platform owns its technology landscape and standardized interaction — Strategy and road map. The joint vision mechanisms with other platforms (for example, combines business and technology outcomes leveraging APIs). It also has an inherent to deliver end-to-end value. Close alignment objective to modernize its technology. between the business unit and the technology group on performance objectives and agenda Platform categories unites all members of the platform around a In most cases, a platform can be thought of shared strategic vision, with a road map for as a nimble fintech group in one of three main executing priorities that balance change and categories: business platforms, enterprise resiliency. platforms, or enabling platforms (Exhibit 1). Exhibit 1 The platforms crucial to a bank’s success can be grouped into three categories. Business Platforms directly aligned to a business unit to deliver business and technology Example platforms platforms outcomes (eg, revenue growth, profitability) Consumer lending Cards Wealth management Enterprise Platforms aligned to multiple Enterprise Act as service providers Core banking platforms business units to deliver shared largely enabling other Payments services platforms Analytics and data outcomes across units Enterprise Act as business owners Finance Tech assets providing similar support delivering services Risk services aggregated to create units across the enterprise HR a center of excellence Enabling Platforms not aligned to a business unit Enterprise architecture platforms • Provide scale benefits through consolidation IT infrastructure • Safeguard the bank by defining guardrails Cybersecurity • Enable business and enterprise platforms to deliver business outcomes Source: McKinsey & Company Platform operating model for the AI bank of the future 54

Business platforms are aligned to business microservices, and standardized data access and units and deliver joint business-and-technology governance. Other enterprise platforms aggregate outcomes. As an example, a business platform for support functions such as finance, risk, and human consumer lending would include several cross- resources within a center of excellence. functional teams, each of which owns front-end technology assets and includes business teams for Enabling platforms support other platforms by a specific function or service area. ensuring that technical functionality is delivered quickly and securely at scale. This approach has One team might focus on preapproval and new- proven effective at maximizing scale benefits customer acquisition, with responsibility for while protecting the enterprise with standardized next-generation credit-scoring models using processes. Examples of enabling platforms include traditional data sources (such as credit bureau core technology infrastructure, DevOps tools and reports and internal transaction histories) capabilities, and cybersecurity. and nontraditional sources (including, upon the customer’s permission, tax returns, online Implementing a platform operating presence, partner ecosystem transactions, and model requires five main building more). Another team often takes responsibility blocks for loan underwriting, determining credit limits for individual accounts in accordance with enterprise The distinct advantage of a platform operating risk policy. A third team might focus on consumer model is the foundation it provides for business- insights and personalized messaging, including and-technology partnerships focused on delivering machine-learning decision models and marketing leading-edge AI-enabled solutions (Exhibit 2). As technology (“martech”) tools to deliver intelligent they begin planning the transition from hierarchical credit offers to new and existing customers. The silos to a network of horizontally interconnected customization team owns the design, development, platforms, bank leaders should focus on five main and management of product configurations building blocks: agile ways of working, remote to ensure that each solution addresses the collaboration, modern talent strategy, culture and customer’s precise needs. capabilities, and architectural guardrails. The value and efficiency that can be derived from platform Other teams focus on services and capabilities to operating models are possible only if organizations support external developers and other technology design their operating model to enable these five partners, including, for example, partner elements. Once they have established their vision onboarding and sandbox management and APIs of the new management approach, they should supporting customer journeys and experiences develop a road map for implementing the platform (managing standards and documentation through model. development hubs or platforms). Still other teams support the consumer lending platform by 1. Agile ways of working managing technology—for example, provisioning By extending the platform structure to all groups, of cloud infrastructure. an organization gains the ability to quickly redirect their people and priorities toward value-creating Enterprise platforms enable diverse business opportunities.² For this model to work, however, platforms by providing shared services such banks need to develop agile mindsets within each as vendor management and procurement, team and equip team members with agile ways standardization of cloud and DevSecOps tooling,¹ of working, such as rapid decision and learning build-to-stock process APIs and reusable cycles, breaking initiatives into small units of 1DevSecOps tools integrate security measures with DevOps processes. 2Wouter Aghina, Christopher Handscomb, Jesper Ludolph, Daniel Rona, and Dave West, “Enterprise agility: Buzz or business impact?,” March 2020, McKinsey.com. 55 Platform operating model for the AI bank of the future

Exhibit 2 The platform operating model for consumer lending captures the agility of a fintech and scale of a large enterprise. fintech and scale of a large enterprise. Illustrative future state of consumer lending platform Key shifts Standalone micro-services Consumer lending platform Wealth Cards Business platforms completely own • External partnerships front-end technology assets and a • Credit underwriting cross-functional team for • Consumer lending policies • Product definition — Onboarding and sandbox for partners — Experience APIs controlled via contracts Business Experience APIs for digital commerce — Consumer-lending specific platforms ecosystems microservices for bureau, credit scoring Lending (new build) Credit limit — Provisioning of cloud infrastructure — Consumer-lending specific AA models Bureau check (from IT) — Product configuration Credit model Loan cross-sell model — Credit underwriting adjustments (in line with overarching risk policy) Enterprise Core banking services Enterprise platforms provide core shared platforms services for business platforms, including: Payments — Standardization of cloud/DevSecOps Analytics and data (data lake, standards, analytical tools, tooling governance) — Standardized data access and governance Enabling Enterprise architecture Only few core technology foundations like platforms Delivery enablement (DevOps) enterprise infrastructure and DevOps Cybersecurity and technology risk would be common across the bank Infrastructure/site reliability engineering Cloud infrastructure and applications work, piloting new products to get user input, and diverse groups have already achieved a degree of rapidly testing operational effectiveness before organizational and operational flexibility, the time scaling.³ This methodology, when deployed across may be right for an end-to-end transformation the organization, underpins a new corporate program that “flips” the organization to agile. culture that enables fast communication and collaboration within and among platforms. It gives Each platform consists of one or multiple squads or the organization a strong and stable backbone for pods combining IT, design, and customer-journey developing and scaling dynamic capabilities. experts, among others (up to nine people). Banks should also create “chapters” as cross-squad The starting point depends on where the bank is groups of employees with similar functional in its technology transformation. Some may set up competencies to ensure growth of expertise and an agile pilot within a platform and gradually train cross-training of colleagues across technologies. other groups in the new practices. For banks where In some cases, a bank will need to create new roles, 3Anusha Dhasarathy, Isha Gill, Naufal Khan, Sriram Sekar, and Steve Van Kuiken, “How to become tech-forward: A technology-transformation approach that works,” November 2020, McKinsey.com. Platform operating model for the AI bank of the future 56

such as tribe leaders and agile coaches. It is also accessibility, software accessibility, and tooling— crucial to adopt a performance-management model to support secure and efficient remote work. that aligns all individuals with team goals. For example, setting clear decision-making The agile way of working is a means to an end, not and escalation paths is essential to maintain an end in it itself. As banks begin to implement a a fast cadence. Shared workflows, roles, and platform operating model, it is crucial that they responsibilities help move work through the set a North Star, not only to unite people around pipeline for even the most complex and highly business goals but also to offer them a sense of interactive jobs. meaning and purpose within society. Shared values reinforce team spirit and—when combined with Setting up a single source of truth or single opportunities to learn, experiment, and make a backlog of work also helps keep different difference for customers—strengthen employee platforms aware of interdependencies. What is engagement. This stronger employee engagement more, banks can and should ensure the security can be measured in, for example, productivity and of remote working arrangements by leveraging loyalty and can indicate how well an organization specialized technology for managing remote has embraced the agile transformation. access. Areas subject to management may include data retrieval (role-based access to 2. Remote collaboration data, restrictions in downloading sensitive data, For a variety of reasons, including geographic restriction of all data copying even on encrypted distribution, work-from-home policies, travel removable hard drives), sophisticated detection restrictions, and other disruptions due to COVID- (tracking and monitoring mechanisms to detect 19, banks have moved to a fully or partially data breach), and governance procedures to remote model. The sharp decline in co-location review breaches and enforce corrective actions. has put pressure on organizations to improve collaboration and consistency in ways of working. Banks should also set up mechanisms to address Given the expectation that a significant share of both interaction and security criteria. These bank employees may not return to shared work mechanisms are particularly crucial for remote- environments,⁴ banks need to develop mechanisms working arrangements, which are increasingly to support effective collaboration—and thus reduce important to top talent in technology-intensive errors—in distributed environments. industries, including financial services. Indeed, banks need to revisit agile teams after an 3. Modern talent strategy abrupt shift to remote models⁵ and consider the A modern talent strategy for an AI bank is not only types of work to be done remotely according to how about the commitment and capability to hire the well interaction models and system readiness can best engineering talent or the best business talent. be adapted. Two criteria are key for determining The AI-bank operating model also requires leaders which roles can function effectively in remote to rethink their strategy for hiring and retaining work arrangements. First is the required level of top talent in a world with blurring lines between human interaction, such as the degree of real-time business, IT, and digital expertise. Leaders must collaboration and creative work among groups form a detailed picture of the diverse skills and of people and the degree to which work can be expertise required to deliver business-technology segmented and individualized. Second is bank solutions. Reskilling is equally critical to building systems’ readiness—particularly in terms of data teams with the right mix of talent. 4Susan Lund, Anu Madgavkar, James Manyika, and Sven Smit, “What’s next for remote work: An analysis of 2,000 tasks, 800 jobs, and nine countries,” McKinsey Global Institute, November 2020. 5Santiago Comella-Dorda, Lavkesh Garg, Suman Thareja, and Belkis Vasquez-McCall, “Revisiting agile teams after an abrupt shift to remote,” April 2020, McKinsey.com. 57 Platform operating model for the AI bank of the future

This strategy focuses on attracting digital talent want to stay and grow within the organization and and requires that leaders understand the unique so all employees see and embrace the change and needs of digital talent. It employs a diversified invest in upgrading their skills. In short, banks need approach to recruiting: engaging with technologist to become great engineering organizations.⁶ communities, sponsoring hackathons to scout talent, and ensuring that recruiters have experience 4. Culture and capabilities in technology. The best technical talent has a As banks build sophisticated technical solutions, disproportionately higher impact, so the ability to they also need to develop a culture suited to the attract and develop superior candidates is crucial. In experts building these solutions. Organizations a similar vein, leading tech organizations enlist their need to manage culture and capabilities to create top performers in the recruiting effort. a virtuous circle that attracts talent, sparks innovation, and creates impact. This underscores Furthermore, banks need to improve retention the importance of talent and culture in tech-enabled and reskilling. Reskilling may involve charting a transformations,⁷ including AI-bank transformations. clear career development path for digital talent, creating an environment that prioritizes and rewards For the platform operating model to work, leaders learning, and rewarding deep expertise over need to steer their organizations to focus on fungible skill sets. There is also opportunity to build the end user, collaborate across silos, and capability-development programs that help reskill foster experimentation. Establishing this digital nontechnical colleagues as technologists. Finally, culture across the bank involves addressing four so that attracting and developing digital talent can dimensions of culture: understanding/conviction, produce the desired results, banks need a clear reinforcement, reskilling, and interaction. strategy for retaining this talent, such as providing flexible and collaborative ways of working and First, understanding and conviction follow largely empowering digital talent to implement change. from the bank’s leadership, expressed through role modeling and encouraging desired behaviors, To develop a comprehensive talent strategy, an including continuous learning, knowledge-sharing, AI bank would first review existing initiatives, the and interdisciplinary collaboration. For example, if structure and makeup of each platform, and the a top team visibly takes part in upskilling programs technical talent required to execute the strategy. for AI and machine learning, this demonstrates to The second step is to build from the ground up all in the organization the importance of automation a model of talent required for the next stage of and evidence-based decision making to all parts growth, including both existing and future initiatives. of the business. Another approach is to support Next, it is important to create a set of talent technology start-ups by giving them access to interventions that can tap into existing talent within nonsensitive code and shareable data to build their the organization, developing an “ecosystem” of own “open solutions” related to AI banking. partners (vendors, developers, gig workers, remote talent, and others) and using hiring mechanisms, The second is to reinforce new practices with formal including the acquisition of smaller companies mechanisms, so that the structures, processes, and and start-ups, to establish platforms requiring systems of the AI bank become embedded within skills beyond the traditional scope of the bank’s the culture. For example, banks might consider roles and capabilities. Finally, banks have to make organizing institution-wide innovation challenges themselves externally appealing to fresh tech talent or inviting managers to daily huddles where they and internally exciting for their people. This means actively work with the centers of excellence to solve transforming themselves so top technical talent problems and own outcomes. 6 Abhishek Chakravarty, Dave Kerr, and Nina Magoc, “10 Principles That Build Great Engineering Organizations,” March 26, 2021, medium.com. 7Reed Doucette and John Parsons, “The importance of talent and culture in tech-enabled transformations,” February 2020, McKinsey.com. Platform operating model for the AI bank of the future 58

Third, leaders need to ensure that every individual individual platforms. These various responsibilities has access to the skills they require to be effective. are formally documented and communicated widely. One way to do this is by developing entirely new Without such guardrails, inefficiencies would tools and technology using in-house open-source multiply. systems. Another is to ensure transparency by setting up digital wikis that anyone can use to These architectural guidelines should focus on access knowledge. Organizations can also learn strategic activities rather than operational tasks, from others by sending employees on “innovation which are subject to the discretion of the platform. tours” or actively encouraging and sponsoring This requires significant time upfront for strategic attendance at high-quality conferences. planning, and each platform must stay alert to new value-creation opportunities related to its mandated Finally, leaders should model various approaches strategic objectives. to interaction. Banks can visibly change the ways managers interact with teams, such as by Further, platform owners can evaluate the moving from meetings to offline asynchronous effectiveness of these guardrails by tracking the communications using highly collaborative tooling. number of business capabilities in accordance with Leaders can also use symbols in remote and these guardrails, rather than simply counting the in-person meetings to emphasize enterprise values various technology applications found within the such as customer centricity. At a leading bank, for organization. example, every meeting has an empty chair to remind participants of the customer for whom they Mapping the operating model of a are building solutions. financial-services organization 5. Architectural guardrails A large global or regional AI bank implementing a Each platform is responsible for its own technology platform-based operating model would typically landscape, but standardized mechanisms for have 20 to 40 platforms, each focused on a specific interaction among platforms should be jointly type or set of services, such as payments, lending, designed across all platforms. It is important, infrastructure, or cybersecurity (Exhibit 3). As noted therefore, to ensure that architectural guardrails above, these platforms are often grouped into one are observed so that each platform can easily of three areas. interact with others. These guardrails should not be perceived as restricting platforms from developing — Business platforms typically include a consumer and improving their own technology and technical platform, which is linked to channels (digital, decisions. branch) and products (wealth, consumer) as well as customer relationship management As each platform is free to build the and analytics; a corporate platform, which technology elements required to deliver on its spans channels and products (transaction mandated business goals, there is potential banking, lending) and relationship management for miscommunication among platforms. For (corporate servicing); and a global-markets example, instead of developing its own interest platform, which covers channels, products, and rate calculation, a consumer lending platform global market operations, as well as market and would leverage a single, standard calculation via credit risks. an API. With no guardrails in place, there would be significant inefficiency, because efforts would — Enterprise platforms provide shared services be duplicated in some areas and tasks would across different business platforms across the be unfinished in others. By contrast, guardrails enterprise on administrative elements such as support efficient management and operation of the customer servicing; employee services; finance; overall IT landscape, with responsibility for various HR; risk, legal, and compliance; and technology elements of the enterprise architecture delegated to platforms usable by business platforms such 59 Platform operating model for the AI bank of the future

Exhibit 3 If implemented across the bank, the platform operating model can enable each group to optimize performance. Business Consumer platforms Corporate platforms Global Markets platforms Digital and assisted digital channels Digital and assisted digital channels Digital channels Branches and self-service banking Trading Wealth products Transaction banking (securities and Product control Consumer products fiduciary services, trade finance and Global markets operations Consumer banking operations cash management) Market risk Customer marketing and analytics Lending and other products Credit risk Corporate servicing and operations Sales and analytics Customer marketing and analytics Enterprise Payments utility (fulfilment and settlement, Finance and HR (recruiting, talent management, platforms payment interfaces, remittances) HR policies, accounting) Customer servicing (reconciliation, digital servicing) Risk (credit, market, operational, and liquidity risk) Analytics and data (data lake, standards, analytical Compliance tools, governance) Group services (eg, strategic vendor management, Employee services (intranet, facilities booking, real estate, project management office) video conferencing, end-user computing) Core banking Enabling Enterprise architecture (application/data/infrastructure architecture, API standards) platforms Delivery enablement/ITSS (DevOps, agile, test automation, service monitoring) Access management (eg, single sign-on, authentication, token management) Cybersecurity and technology risk Infrastructure/site reliability engineering API management (tech and operations for all APIs) Cloud infrastructure and applications as payment infrastructure, cloud infrastructure, The platform model can help data, and API management. organizations seize new opportunities — Enabling platforms support business and Executing on a platform operating model is arduous. enterprise platforms to deliver technical However, when done correctly, it has the potential functionality quickly. These platforms include to deliver four main benefits to all stakeholders: enterprise architecture, delivery enablement, value-oriented business-technology partnerships, access and authentication management, stronger performance (speed, efficiency, and cybersecurity, and infrastructure/site reliability productivity), transparency, and a future-ready engineering (SRE). business model. Platform operating model for the AI bank of the future 60

The collaborative framework of the platform model domain expertise and agile skills for collaboration brings business and technology leaders together and timely delivery. as co-owners in creating value for the enterprise. Joint owners of business-facing platforms share In addition to the emphasis on interdisciplinary accountability for outcomes, merging business collaboration, the platform model is designed knowledge of market opportunities with expert to increase transparency, accountability, and insight into how technological advances can knowledge sharing to the fullest extent possible. enhance customer experiences. The leader of the Transparency should be high not only so employees platform facilitates the interaction of business can clearly identify the services available from and technology owners in determining the right each platform but also to support independent balance between run-the-bank and change-the- benchmarking of team performance and bank initiatives. All members of a particular team identification of best practices. Each platform are unified in delivering a solution (just as those should also be clear about how it prioritizes work, of the entire “tribe” of a platform are focused on a tracks initiatives in the pipeline, and manages the service line) in order to create value in alignment backlog. with enterprise strategic objectives. This unity is reinforced by the fact that all team members share Finally, shifting to a platform model can help in performance metrics for both business and an organization future-proof its business technology outcomes, including impact on users model because each platform is incentivized to (internal and external), on-time delivery of solutions, continuously improve on its technology landscape. customer and employee satisfaction ratings, Within a culture of continuous learning, team and more. members are accustomed to change and adept at finding the best response to fast-evolving The platform approach can strengthen an circumstances. Interdisciplinary initiatives led by organization’s performance in terms of speed, business-technology co-owners strengthen a efficiency, and productivity when each platform is team’s capacity to anticipate and consider potential large enough to address a set of use cases crucial challenges and opportunities before they appear to realizing the business model of the enterprise on the horizon. Enterprise-wide standards, rigorous but small enough to keep the team agile. Each documentation of processes, and consistent team enjoys a degree of autonomy, with a budget cataloging of technology assets enable teams to and mandate to experiment and discover the best apply best practices as they develop and implement way to maximize value within a discrete domain in new solutions. alignment with predefined guardrails (for instance, finance, risk, compliance) without having to wait By underpinning business-technology for approvals from finance and allocations from co-ownership of solutions delivery and value IT and human resources. This autonomy speeds creation, the platform operating model offers up decision making, innovation, and solution banks an opportunity to maximize the impact of delivery. The use of automated tools, enterprise their technology capabilities in ways that count for standards, and agile patterns of communication and customers. The implementation of the platform collaboration increases efficiency in two ways. First, model begins logically with the formation of joint this approach minimizes duplication of effort by business-and-technology teams focused on the documenting repeatable processes and cataloging design, development, and implementation at scale technology tools and analytical models available for of new AI-bank innovations, always striving toward deployment in diverse contexts. Second, it allows a more intelligent value proposition and smarter individuals to access data (according to clearly experiences and servicing. Further, the creation defined need-to-know criteria) and advanced of cross-functional platforms is also an excellent analytical tools to extract insights to augment their impact. Over time, persistent agile teams build their 61 Platform operating model for the AI bank of the future

approach to increase business–technology these elements together is a powerful mechanism collaboration, developing an IT operating model to optimize the full capability stack, from core that generates immediate and tangible business technology and data infrastructure to AI-powered value and moves the full organization, not just decision making and reimagined customer technology, to an agile way of working. However, engagement. The platform operating model to derive maximum value from platforms and the ensures that these layers run in sync to spur the people who make up these platforms requires new growth of an AI bank of the future. skills, mindsets, and ways of working. Bringing all Brant Carson is a partner in McKinsey’s Sydney office; Abhishek Chakravarty is an associate partner in the Singapore office, where Kristy Koh is an associate partner; and Renny Thomas is a senior partner in the Mumbai office. Platform operating model for the AI bank of the future 62

Contacts For more information about this report, please contact: Renny Thomas Senior Partner, Mumbai [email protected] Akshat Agarwal Xiang Ji Associate Partner, Bengaluru Partner, Shenzhen [email protected] [email protected] Anubhav Bhattacharjee Dave Kerr Solution Leader, Mumbai Senior Expert, Singapore [email protected] [email protected] Suparna Biswas Kristy Koh Partner, Mumbai Associate Partner, Singapore [email protected] [email protected] Sven Blumberg Milan Mitra Senior Partner, Düsseldorf Associate Partner, Bengaluru [email protected] [email protected] Brant Carson Kayvaun Rowshankish Partner, Sydney Partner, New York [email protected] [email protected] Abhishek Chakravarty Anushi Shah Associate Partner, Singapore Consultant, Mumbai [email protected] [email protected] Violet Chung Shwaitang Singh Partner, Hong Kong Partner, Mumbai [email protected] [email protected] Malcolm Gomes Charu Singhal Partner, Bengaluru Consultant, Mumbai [email protected] [email protected] Rich Isenberg Yihong Wu Partner, Atlanta Practice Manager, Hong Kong [email protected] [email protected] Editor: John Crofoot Design and production: Matt Cooke, Kate McCarthy, Paul Feldman A special thank you to: Arihant Kothari, Amit Gupta, Himanshu Satija, Jinita Shroff, Vineet Rawat 63 Building the AI bank of the future

May 2021 Copyright © McKinsey & Company www.mckinsey.com @McKinsey @McKinsey

Global Banking & Securities Beyond digital transformations: Modernizing core technology for the AI bank of the future For artificial intelligence to deliver value across the organization, banks need core technology that is scalable, resilient, and adaptable. Building that requires changes in six key areas. This article was a collaborative effort by Sven Blumberg, Rich Isenberg, Dave Kerr, Milan Mitra, and Renny Thomas. © Getty Images April 2021

An artificial-intelligence (AI) bank leapfrogs An AI-first model places demands on the competition by organizing talent, technology, a bank’s core technology and ways of working around an AI-first vision for empowering customers with intelligent value Across industries, many organizations have propositions delivered through compelling struggled to keep pace with the demand for journeys and experiences. Making this vision digitization, especially as consumers accelerated a reality requires capabilities in four areas: their adoption of digital channels for daily an engagement layer, decisioning layer, core transactions during the COVID-19 crisis.³ Even technology layer, and platform operating model. before that, however, the financial-services industry has historically had mixed success Previous articles in this series have explored in technology. Institutions that were early the first two areas. The capabilities of the adopters and innovators in technology have reimagined engagement layer¹ enable the AI built up a complex landscape of technical assets bank to deliver highly personalized seamless over decades and accumulated significant journeys across bank channels and within technical debt. Some institutions have tackled partner ecosystems. The capabilities of the this challenge; many are behind the curve. AI-powered decisioning layer² transform Meanwhile, alongside the incumbents, an customer insights into messages and offers extremely active fintech industry has been tailored to address a customer’s unique needs. constantly innovating and raising the bar. The current article identifies capabilities needed in the third area, the core technology and data Financial institutions that have shifted from infrastructure of the modern capability stack. being intensive consumers of technology to making AI and analytics a core capability are Deploying AI capabilities across the organization finding it easier to shift into the real-time and requires a scalable, resilient, and adaptable consumer-centric ecosystem. As AI technologies set of core-technology components. When play an increasingly central role in creating value implemented successfully, this foundational for banks and their customers, financial-services layer can enable a bank to accelerate technology organizations need to reinvent themselves innovations, improve the quality and reliability as technology-forward institutions, so they of operations, reduce operating costs, and can deliver customized products and highly strengthen customer engagement. personalized services at scale in near real time. We begin by summarizing the primary demands At many institutions, standard practices now banking leaders should consider as they plan include omnichannel engagement, the use of an enterprise-wide initiative to modernize APIs to support increased real-time information core technology, data management, and the exchange across systems, and the use of big underlying infrastructure. Next, we examine data analytics to improve credit underwriting, the key transformations required to modernize evaluate product usage, and prioritize the core technology and data infrastructure. opportunities for deepening relationships. As We conclude by sharing 12 actions technology financial-services organizations continue leaders should consider taking to ensure the to mature, the increasing demands on the transformation creates value for customers and technology infrastructure to support more the bank. complex use cases involving analytics and real- 1Violet Chung, Malcolm Gomes, Sailee Rane, Shwaitang Singh, and Renny Thomas, “Reimagining customer engagement for the AI bank of the future,” McKinsey.com, October 13, 2020. 2Akshat Agarwal, Charu Singhal, and Renny Thomas, “AI-powered decision making for the bank of the future,” McKinsey.com, March 23, 2021. 3Tamara Charm, Anne Grimmelt, Hyunjin Kim, Kelsey Robinson, Nancy Lu, Mayank, Mianne Ortega, Yvonne Staack, and Naomi Yamakawa, “Consumer sentiment and behavior continue to reflect the uncertainty of the COVID-19 crisis,” October 2020, McKinsey.com. 2 Beyond digital transformations: Modernizing core technology for the AI bank of the future

time insights are pushing firms to reexamine demand and capacity to meet strategic and near- their overall technology function. Once they have term priorities, and a well-defined mechanism to committed to modernizing the core technology coordinate “change the bank” and “run the bank” and data infrastructure underpinning the initiatives according to their potential to engagement and decision-making layers of the generate value. capability stack, banks should organize their transformation around six crucial demands: Faster time to market requires efficient and technology strategy, superior experiences, repeatable development and testing practices scalable data and analytics platforms, scalable coupled with robust platforms and productivity- hybrid infrastructure, configurable product measurement tools. Aligning demand and capacity processors, and cybersecurity strategy (Exhibit 1). according to strategic priorities works on two levels. On one level, banks need to ensure that Robust strategy for building technology execution, infrastructure, and support capacity are capabilities optimized to ensure constant operation of all use Before embarking on a fundamental cases and journeys. On the other, with constant transformation of core technology and data uptime assured, work should be organized and infrastructure, financial-services organizations scheduled to expedite projects having the greatest should craft a detailed strategy for building impact on value. Finally, financial institutions an AI-first value proposition. They should also should establish clear mechanisms for setting develop a road map for the transformation, priorities and ensuring that each use case is focusing on three dimensions of value creation: designed and built to generate a return exceeding faster time to market with efficient governance capital investments and operating costs. and productivity tracking, clear alignment of Exhibit 1 The AI-bank transformation places several crucial demands on core technology and data infrastructure. Robust strategy for building Superior omnichannel journeys Modern, scalable platform for technology capabilities and customer experiences data and analytics Scalable hybrid infrastructure Highly configurable and Secure and robust strategy for the cloud scalable core product processors perimeter for access Beyond digital transformations: Modernizing core technology for the AI bank of the future 3

Superior omnichannel journeys and customer Modern, scalable platform for data and experiences analytics Building journeys that excite customers with Delivering highly personalized offers in near their speed, intuitiveness, efficiency, and impact real time requires AI-powered decision-making typically involves various applications spanning capabilities underpinned by robust data multiple bank and nonbank systems, all linked assets. What is more, the at-scale development together by a series of APIs and integrations. of machine-learning (ML) models that are This complex information exchange enables the context aware in real time requires automated organization to ingest valuable data from diverse DevSecOps⁴ and machine-learning ops (MLOps) sources to produce highly personalized messages tools to enable secure and compliant continuous and offers that speak directly to the customer integration (CI) and continuous deployment (CD). in near real time. In addition to a standardized This entails complex orchestration across source approach to managing APIs, banks should develop systems, data platforms, and data sciences a clear mechanism to integrate across channels, to enable lab experimentation and factory core systems, and external interfaces while production. This is particularly complex in a highly managing changes across multiple dependent regulated environment where the involvement of systems. They should bear in mind, for example, security, audit, risk, and other functions is crucial that introducing a change in an existing digital in many stages of the process. channel could potentially entail changes not only across the front end but also across multiple The incorporation of feedback loops with channel interfacing systems, core product processors, and systems enables models to evaluate the output analytics layers. performance and make automated adjustments to increase the effectiveness of personalized A focus on journeys and user experience also messages, so the organization can generate benefits back-office and operations teams. New personalized offers nearly instantaneously. For products are increasingly automated at the back example, in the case of location-based offers end, freeing staff to focus on genuinely exceptional for adjacent products, an organization must be scenarios and differentiating activities, rather than able to overlay in real time customer location repetitive low-value activities. and preferences (as reflected in previous transactions) with predefined offers from nearby Finally, to ensure maximum value, use cases and participating merchants. capabilities should be designed as “enterprise products” to be reused in other areas. For example, Scalable hybrid infrastructure utilizing the cloud the deployment of microservices handling discrete With the continued expansion of customer tasks like document collection and ID verification engagement across bank and nonbank platforms, can ensure consistency in the way things are financial institutions need to create hyperscalable done across the organization. APIs should also infrastructure to process high-volume be documented and catalogued for reuse. APIs transactions in milliseconds. This capability that are domain- or product-centric (for example, is made possible, in part, by infrastructure enabling the retrieval of customer details from a as code, automated server provisioning, and single customer store) have higher reusability and robust automated configuration management take an enterprise-level view of the capability, as processes, which together solve the problem of compared with a journey-centric API design—for “snowflake” configurations resulting from organic example, one where an API supports retrieval of and complex linkages and changes that have customer details for a specific mobile journey. accumulated over time. 4DevSecOps tools support the integration of “development, security, infrastructure, and operations at every stage in the product’s life cycle, from planning and design to ongoing use and support.” See Santiago Comella-Dorda, James Kaplan, Ling Lau, and Nick McNamara, “Agile, reliable, secure, compliant IT: Fulfilling the promise of DevSecOps,” May 2020, McKinsey.com. 4 Beyond digital transformations: Modernizing core technology for the AI bank of the future

Hosting these environments on a distributed- as protection against vulnerabilities within network cloud environment allows a balance applications, operating systems, hardware, between paid-up-front baseline storage and and networks. Financial institutions should computing capacity, on the one hand, and, on the also implement appropriate measures to other, elastic on-demand surge capacity without secure the perimeter and control access disruptions to service. Self-monitoring and to various systems and applications within preventive maintenance also are automated, and the organization’s infrastructure footprint, disaster recovery and resiliency measures run in including private and public cloud servers the background to ensure constant uptime even and on-premises data centers. For example, if incidents evade automated self-repair and transferring workloads from traditional require manual intervention. As a result, the risk on-premises infrastructure to public cloud of disruption to critical operations is minimized, requires careful measures to protect customer and customer-facing applications run with high data, along with a robust strategy for detecting availability and responsiveness. The combination and remediating potential threats and of on-premises and cloud-based infrastructure vulnerabilities. is increasingly relevant in high-volume and high- frequency areas such as payments processing, The “classical” approaches of securing the core banking platforms, and customer perimeter should be coupled with more modern onboarding systems. Making workloads “cloud approaches to limit the impact of intrusions or native” and portable allows the work to be moved reduce the “blast radius.” Again, AI has a part to the most appropriate platform. to play here, given the advent of increasingly sophisticated network intrusion detection, Highly configurable and scalable core product anomaly detection, and even forensics during processors postmortems of security incidents. To sustain a leading-edge value proposition founded upon AI and ML capabilities, banks Start the transformation by must continually evaluate their core products prioritizing key changes and identify opportunities for innovations and customizations. Combined with deep To meet these demands, financial institutions understanding of customer needs, enabled will need to transition from a legacy by advanced analytics, an organization can architecture and operating model to an anticipate emerging customer requests and automation and cloud-first strategy. Building design distinctive products accordingly. The the core technology and data capabilities upon need for real-time reconciliation and round- a highly automated, hybrid-cloud infrastructure the-clock transaction processing also emerges can enable the AI bank to scale rapidly as a key competitive advantage for financial and efficiently as it gains competitive and institutions. For example, with the advent differentiating capabilities. of next-generation core banking platforms, organizations can now develop products that are The AI-bank capability stack combines core built for scale and can be readily configured to systems and AI-and-analytics capabilities in meet specific customer expectations.⁵ a unified architecture designed for maximal automation, security, and scalability. Getting Secure and robust perimeter for access to this target state requires a series of complex It is crucial to ensure that the organization initiatives to transform the organization’s core maintains an appropriate cybersecurity posture technology and data infrastructure. These across the entire technology infrastructure initiatives focus on several key areas: tech- 5Xavier Lhuer, Phil Tuddenham, Sandhosh Kumar, and Brian Ledbetter, “Next-generation core banking platforms: A golden ticket?” August 2019, McKinsey.com. Beyond digital transformations: Modernizing core technology for the AI bank of the future 5

forward strategy, modern API and streaming time-right releases. Organizations should also architecture, core processors and systems, data adopt enterprise agile practices for high-velocity management, intelligent infrastructure, and engineering teams, with integrated cross-functional cybersecurity and control tower (Exhibit 2). teams of business, technology, and functional experts, and external partners using modern Tech-forward strategy approaches to software development, testing, Banks should begin this far-reaching initiative by release, and support cycles. In addition, efficient translating the AI-first vision into an enterprise management of the full stack requires governance strategy that merges technology with business, of the technology function through a standardized funding investments in innovation with the returns set of metrics, along with ongoing tracking of on incremental changes in technology.⁶ Business uptime and health for each component of the stack. and technology collaborate as co-owners in designing and managing operating models Modern API and streaming architecture and outcomes. This “tech-forward” mindset Next, banks should integrate internal and external thrives in interdisciplinary teams focused on systems to support seamless customer journeys innovation and led by skilled engineering talent across internal platforms, partner ecosystems, leveraging modern tools and practices for first- and numerous external interfaces. This requires 6 The technology transformation described in this and other articles in our series on the AI bank of the future aligns broadly with our colleagues’ discussion of the tech-forward approach, which applies across industries. See Anusha Dhasarathy, Isha Gill, Naufal Khan, Sriram Sekar, and Steve Van Kuiken, “How to become tech-forward: A technology-transformation approach that works,” November 2020, McKinsey.com. Exhibit 2 Building a modern core technology and data infrastructure entails changes in several key areas. Channels and digital-journey integrations Tech-forward Modern APIs and Data management strategy streaming architecture for the AI world Core processors and systems Intelligent infrastructure Cybersecurity and control tower 6 Beyond digital transformations: Modernizing core technology for the AI bank of the future

a robust, scalable, and standardized approach It is crucial to establish a modern data and to building and hosting integrations and APIs. analytics platform to fuel the real-time ML The APIs, in turn, should be rigorously tested for models of the decision-making layer. The performance and developed using agile release analytical insights generated by these models principles. When a well-defined stock of APIs-as- are deployed through martech tools to craft the products are orchestrating flows across systems, intelligent offers and smart experiences that set product innovations can advance from concept an AI bank apart from traditional incumbents. In to production and deployment of minimum viable order to support superior omnichannel customer product within 30 to 60 days. journeys and seamless integration with partner ecosystems, the data platform must be capable To complement a robust API strategy, technology of ingesting, analyzing, and deploying vast leaders should also consider establishing a amounts of data in near real time. high-speed data-streaming channel to enable standardized asynchronous data transfer across The data platform should also provide scalable the enterprise in real time. workbenches with AI and data-science capabilities to lab and factory teams. These Core processors and systems workbenches enable teams to access relevant With the right architecture in place, banks can data sets as they develop models and deploy shift away from traditional, complex, and tightly insights in product iterations. The infrastructure intertwined core systems to lightweight and should also support the development of ML highly configurable core product processors models through automated and repeatable and workflows. These processors are also processes. complemented by “microservices,” or discrete applications (such as for payments, card If an organization allows interdisciplinary teams accounts, or loans) that “externalize” the logic across the enterprise to search and extract within traditional core platforms. data held on the platform, these teams can optimize their data consumption according to The transition to lightweight core processors and customer needs and market opportunities. It systems hosted on scalable, modular, and lean is essential to enable data-science teams with platforms exposed as APIs supports, for example, appropriate tooling and access to scalable real-time reconciliation and allows changes to be computing power so that they may experiment made in live systems with zero downtime. Use of and innovate. Underpinning these actions, modern cloud-based infrastructure to host such appropriate technical documentation and platforms also makes it easier to scale up. cataloging of assets (for example, APIs, ML If successfully implemented, a lightweight models, data dictionary, DevOps and MLOps processor platform can enable an organization tools) ensure proper governance and access to advance from new-product concept to launch control. By creating ML models and scorecards in two to three months. This is a significant through a well-defined lab-factory model, advantage against organizations constrained AI-first organizations empower employees to by legacy technology, where launching a new leverage self-serve, real-time data and analytics product or customizing an existing product infrastructure to guide value-based planning can take six months or more. Assembly of new and support daily decision making. off-the-shelf product stacks can also enable innovative new customer propositions, such Intelligent infrastructure as an end-to-end lending journey on a modern Banks then should ensure they have an stack using these principles. effective strategy to modernize infrastructure. For this, they should consider the adoption Data management for the AI world of public cloud to complement the traditional Beyond digital transformations: Modernizing core technology for the AI bank of the future 7

infrastructure in situations where workloads maintained and modernized their legacy platforms. require resiliency, scale, and use of hosted or Various business lines have set up organically managed offerings (such as hosted databases). built platforms upon this foundation, making it Public cloud enables velocity through higher costlier and more and more complex to maintain. levels of automation, templates, and reduction Many organizations have spent billions of dollars of operational risk. When setting up such on multiyear technology initiatives within silos, environments, banks must build upon the only to find that they fail to generate the scale foundational elements of infrastructure benefits required to justify investments. Leaders management, including observability, resiliency, should heed these lessons, adopt a holistic and high availability, as well as a robust perspective, and map priorities according to the configuration strategy. A well-tuned, scalable, end-to-end impact that each step in the technology and load-balanced stack can support response transformation has on the value of the enterprise. times of less than a second while scaling horizontally to cater to variations in transaction If an organization meets the strategic demands volume. outlined at the top of this article, the implementation of modern core technology and data infrastructure Cybersecurity and control tower can yield significant value in the form of faster Finally, institutions should address cybersecurity delivery of changes and improvements, increased and control. This includes setting up a centralized cost efficiency, higher quality of assets, and control tower to monitor data, systems, and stronger customer outcomes. For example, a sound networks across the infrastructure. The scope DevOps and release-management strategy can of responsibility includes ensuring boundary contribute to a 25 to 30 percent increase in capacity security and identifying and rectifying threats creation, a reduction in time to market of 50 to 75 and intrusions. Also crucial is to establish a percent, and more than a 50 percent reduction well-defined set of compliance measures for in failure rates.⁷ In turn, development efforts can security testing and vulnerability scanning improve schedule adherence by 1.5 times and before deploying assets on live systems. These reduce customer defects by 20 to 30 percent measures reduce the risk posed by potential through process automation and agile ways of threat scenarios. working,⁸ and leading organizations have improved issue-resolution time and planning time by between Technology leaders should prioritize 30 and 50 percent.⁹ There are indirect benefits as interconnected capabilities well: by empowering employees with a clear mission, autonomy, and strong focus on customers, agile Given the broad scope of components to be organizations have been able to increase employee transformed, organizations should bear in engagement by 20 to 30 percent, as reflected both mind that optimal outcomes are much likelier in willingness to recommend their workplaces and in when they first establish a holistic strategy for employee-satisfaction surveys.¹0 technology transformation. Unfortunately, not all have found the resources to embrace fully the Technology transformations are fraught with risk, potential offered by the rapid advancement of including delays and cost overruns, and only those AI technologies and the steady rise in customer organizations whose leaders are prepared to expectations. Some financial institutions, commit the energy and capital necessary to carry despite seeing the imperative to change, have through with the comprehensive effort should 7Thomas Delaet and Ling Lau, “DevOps: The key to IT infrastructure agility,” March 2017, McKinsey.com. 8Matt Brown, Ankur Dikshit, Martin Harrysosn, Shivam Srivastava, and Kunal Thanki, “A new management science for technology product delivery,” February 2020, McKinsey.com. 9Wouter Aghina, Christopher Handscomb, Jesper Ludolph, Daniel Rona, and Dave West, “Enterprise agility: Buzz or business impact?” March 2020, McKinsey.com. 10 “Enterprise agility,” March 2020. 8 Beyond digital transformations: Modernizing core technology for the AI bank of the future

embark on the journey. Ultimately, this is a financial-services leaders that are in the process decision not just to survive, but to thrive, and of carrying out such transformations (Exhibit 3): it requires a change in mindset. Specifically, traditional financial institutions will need to 1. Consider the factory model to build at scale. break out of their legacy technology architecture Leverage a factory approach in fast-evolving and explore AI-and-analytics opportunities. and critical areas of the transformation to Should they undertake the challenge and begin enable repeatable execution and development thinking about how best to chart their course to of capabilities within technology teams and to becoming an AI bank, their leaders may consider promote standardization to speed up execution. 12 key insights gleaned from the experience of For example, a core system factory consisting Exhibit 3 Leaders should consider 12 key insights as they embark on the technology- ttrraannssffoorrmmaattiioonn jjoouurrnneeyy.. Tech-forward strategy Consider the factory model to build at scale Consider insourcing differentiating capabilities Modern APIs and Maintain rigorous documentation on integrations streaming architecture Identify an anchor stack but experiment with others Core processors and systems Maintain automation-first and fast-release posture Consider a modern core for high-velocity areas Data management for Adopt a value-centric approach to building data platforms the AI world Set up a lab and factory for analytics Intelligent infrastructure Define the enterprise cloud strategy Establish end-to-end visibility across the stack Cybersecurity and control tower Identify the right perimeter design for the cloud Ensure data security on the cloud Beyond digital transformations: Modernizing core technology for the AI bank of the future 9

of teams, predefined operating procedures, new capabilities while enabling a modular build of and systems to manage, prioritize, and execute financial products. changes across business units can expedite deployment of new solutions significantly. 7. Adopt a value-centric approach to building data platforms. Take advantage of the fact that data 2. Consider insourcing differentiating and analytics platforms evolve over time, and do capabilities. Based on the eventual outcomes not allow teams to be overwhelmed by the rapid desired, build certain differentiating shift of tooling and available technology. We have capabilities in-house, with robust engineering observed that organizations that budget the support, perhaps starting with APIs, anticipated return of change efforts are able to infrastructure, or the data and analytics prioritize use cases that are functionally simple, platform. fit the road map for building the platform in iterations, and realize economic value along 3. Maintain rigorous documentation on the way. integrations. Remember that the development of engagement systems and comprehensive 8. Set up a lab and factory for analytics. Establish changes in core-technology require significant a lab to experiment with tools and platforms for adjustments to integrations, and substandard efficient development in test-and-learn cycles. documentation of the specifications for these Also, build a central factory for producing and integrations often slows the broader initiative deploying analytics use cases at scale on an to transform the bank. individual stack. 4. Identify an anchor stack but experiment 9. Define the enterprise cloud strategy. Create a with others. Emphasize the importance of common strategy across stakeholders to enable standardization for engineering-centric a structured and systematic migration to the development at scale, and build on a single cloud. Cloud adoption poses multiple firsts in stack to support faster change. At the same the enterprise in terms of security perimeters, time, continue experimenting with other stacks change management, and cloud-migration and and stack components for smaller builds in disposition strategy. order to adopt alternative or newer approaches where the incremental benefits are clearly 10. Establish end-to-end visibility across the defined. technology and infrastructure stack. Recognizing that at-scale digital transformations impose 5. Maintain an automation-first and fast- limitations on volume and scale, implement robust release posture. Adopt an automation-first automated tools to observe stack performance and frequent-deployments posture on fast- and to diagnose and resolve issues. evolving applications and stacks. While initial hiccups are not uncommon, release rails 11. Identify the right perimeter design for the cloud. should be hardened over time to speed up time To safeguard against potential malicious attacks to market. Well-defined release management on cloud-based public-facing applications, and deployments are key to execution velocity. design an appropriate network perimeter that Standardizing through DevSecOps typically optimizes the potential attack radius. unlocks productivity gains of as much as 20 to 30 percent. 12. Ensure data security on the cloud. Design robust data-categorization and data-security 6. Consider a modern core for high-velocity safeguards to avoid critical customer-data areas. Consider modern and lightweight combinations and comply with national data- core systems built on scalable and hybrid protection and data-residency laws. infrastructure to enable an efficient rollout of 10 Beyond digital transformations: Modernizing core technology for the AI bank of the future

If banks are to thrive in a world where customer stack also includes a leading-edge operating expectations are increasingly shaped by the model to ensure that all layers work together in AI-and-analytics capabilities of technology unison to deliver intelligent propositions through leaders, they must rebuild their core technology smart servicing and experiences. The AI bank of and data infrastructure to support AI-powered the future requires an agile culture and platform- decision making and reimagined customer oriented operating model that respond promptly engagement. These are the three “technology to emerging opportunities and deliver innovative layers” of the AI-bank capability stack. The full solutions rapidly at scale. The next article in this series examines the crucial elements of the platform operating model. Sven Blumberg is a senior partner in McKinsey’s Dusseldorf office, Rich Isenberg is a partner in the Atlanta office, Dave Kerr is a senior expert in the Singapore office, Milan Mitra is an associate partner in the Bengaluru office, and Renny Thomas is a senior partner in the Mumbai office. The authors would like to thank Brant Carson, Kayvaun Rowshankish, Yihong Wu, and Himanshu Satija for their contributions to this article. Copyright © 2021 McKinsey & Company. All rights reserved. Beyond digital transformations: Modernizing core technology for the AI bank of the future 11

11/24/21, 3:21 PM BaaS: Why you should know about Banking as a Service BaaS: Why you should know about Banking as a Service 22 November 2021 000 The banking sector is going through a period of rapid change. This was true even before the pandemic, which has had a profound effect on what people expect from their banks and banking services, but the seeds of change had taken root long before. Technology and banking are intrinsically linked, and as technology progresses faster and faster, Tier-1 banks are finding it harder to keep pace with new waves of innovation. Consumer and business expectations of banking services are higher than ever, which has fueled the rapid propagation of new providers that artfully blend financial services with modern technology. At the heart of many of these innovative new ventures is open banking, the widespread implementation of which has galvanised the digitisation of the sector. Banking services are now more easily integrated and interconnected between platforms and providers, with a new market emerging for unique financial products and services. Tier 1 banks are not the force driving this digital revolution, but their solidity and status after decades of high street rule cannot be overturned. Many of these new fintechs rely on the support of these institutions to navigate the intricate jungle of banking licensing, regulations and compliance protocols that govern financial services. Matters grow even more complex with the introduction of non-financial companies offering their own integrated products like branded lending services, credit cards and even mortgages as part of their customer journey. The banking landscape today is full of steady giants unable to benefit from the swell of innovation, and agile challengers unable to meet the sector’s difficult requirements. There is a solution to this problem, a means of bridging the divide in such a way that both sides stand to benefit from the opportunity at hand – and that solution comes in the form of Banking-as-a-Service (BaaS). BaaS to the rescue BaaS solutions translate the advantage held by the Tier 1 banks – security, reliability and regulatory licensing – and offers them to smaller businesses through tailored, customisable financial products that arrive ready-made and fully compliant. Financial and non- financial companies alike no longer need to contend with the headache of building safe, functional products from the ground up, nor with the minefield of banking regulations. Take, for example, a small eCommerce retailer that wants to offer split payments at checkout without including 3rdparty options that redirect users away from the site. This retailer would then have to design and build a fully functional credit lending system from scratch, ensuring users’ data privacy and protecting themselves from fraud. Even if they could manage such a feat, they would then need to jump through the many hoops of the bank licensing system, all of which costs time and money. If this retailer approached a BaaS provider instead, they could receive a fully functional lending solution that works out of the box and comes pre-approved and compliant. Many BaaS providers offer white label solutions, meaning users don’t need to dilute their own branding, which in the highly competitive world of eCommerce has incalculable value for small businesses. With the help of the right BaaS provider, these solutions can be adaptable and scalable, growing organically with the business and evolving to meet customer needs at every level – all at high speed and low cost. The demand for these integrated credit services is already out there – just look to the meteoric rise of Buy Now, Pay Later (BNPL) solutions. The sector is growing at a rate of 39% a year, and in the UK alone there are 10 million online shoppers who avoid retailers that don’t offer split payment options at checkout. For small businesses, failing to offer these kinds of services can exclude themselves from a growing proportion of the market – something no SME can afford to do. By opening themselves up to innovative financial services, they open themselves to a whole new audience of customers and can provide them with a better, more inclusive customer experience Working better together It's not only small businesses that stand to benefit from BaaS solutions – the Tier 1 banks do too. While typically slow-moving and highly risk averse, BaaS allow them to connect with smaller businesses that are more agile and better positioned to capitalise on these opportunities, and in turn reap the benefits themselves. BaaS providers act as a conduit between the well-established old guard and the sharp, bright-eyed actors who are driving a market full of innovation, forming lasting and productive partnerships that will reshape banking and financial services as a sector. Partnerships are at the heart of Yobota’s own operations, informing the way we provide our services and how we engage with both sides of the issue. Our core team is made up of ex-bankers and tech experts, providing us with insights from both worlds and helping us tailor solutions that help the old and the new benefit from the opportunities at hand. New ideas promote growth I’ve heard some misguided concerns that the growing popularity of white label BaaS solutions will reduce differentiation within the sector, homogenising banking services and leading to stagnation. This simply makes no sense – white label solutions allow businesses and brands to create products uniquely suited to their own clientele, leading to an outpouring of new, useful services that cater to consumers of all kinds. It’s clear that there is a great opportunity on hand for BaaS providers. Those who can provide the best, most scalable products – products that can match and keep up with companies of all descriptions – are poised to grow in tandem with the new financial services market and redefine what people expect from their banking products, how they spend and how they engage with brands. https://www.finextra.com/blogposting/21290/baas-why-you-should-know-about-banking-as-a-service 1/2