Agentic AI in Banking Will Follow Three Tracks. Fintechs Lead in All of Them
By David Evans, Chief Content Officer at The Financial Brand
Simple Subscribe
Subscribe Now!
More than half of financial services organizations already deploy AI agents to varying degrees, yet adoption follows divergent paths: Incumbent banks concentrate investments on back-office efficiency gains where 68% report substantial returns, while digital-native fintechs aggressively embed agentic capabilities throughout customer-facing experiences and product design.
According to a recent report from Oliver Wyman, this dual-track adoption creates competitive imbalances as three transformative economies emerge: the Assistance Economy where agents deliver complete customer experiences, Adaptive Customer Experiences enabling real-time interface personalization, and Agentic Twins representing customers with delegated authority across financial relationships.
Market leadership will belong to institutions that industrialize agentic AI in all three dimensions as foundational capabilities rather than layering new tools onto legacy workflows.
Need to Know:
- More than 50% of financial services executives report using AI agents to varying degrees across organizations, with specific adoption rates including 57% for customer service applications, 48% for marketing functions, 43% for fraud management, and 40% for productivity and research tasks.
- Yet only 32% of banks realize significant returns from customer-facing AI investments despite 99% prioritizing these initiatives, while 68% agree the most substantial value stems from back-office efficiency gains, a significant misalignment between investment priorities and realized returns.
- A 30% task performance threshold represents current AI model limitations in finance, with model outputs rated better than industry expert outputs across only three-tenths of finance-related tasks on leading benchmarks, necessitating continued human-in-the-loop oversight for complex processes.
The AI Evolution Hits a Critical Inflection Point
Current generative AI tools represent merely the tip of the technological iceberg. The financial services industry approaches fundamental reimagining of service delivery and execution as today’s AI assistants providing help transition to AI agents that autonomously perform tasks. This shift moves beyond isolated point solutions toward systems capable of planning, deciding, and executing complex workflows independently, triggered by single commands rather than requiring continuous human guidance through each step.
Rather than advisory support, the next frontier features autonomous action. AI mortgage agents could instantly compare conveyancers, identify optimal loan rates based on customer financial profiles, coordinate with third parties on client behalf, submit applications, and present validated loan options for final user selection—all from single commands. This operating model builds on dual agent relationships: agent-to-client and agent-to-agent collaboration creating fundamentally new value creation mechanisms.
The agentic era drives transformation not only in customer experience but in operational fundamentals. By leveraging AI agents, banks and fintechs unlock underserved market opportunities including narrow customer segments and persistent advice gaps previously inaccessible or unprofitable. Simultaneously, these capabilities raise bars for acquiring, winning, and critically sustaining customer relationships. Much like how the gig economy emerged from mobile technology and GPS convergence, new economic models will arise around agentic AI through deep embedding within real customer journeys rather than merely layering capabilities onto legacy systems.
Divergent Adoption Strategies Create Competitive Divergence
Banks and fintechs pursue AI adoption along different trajectories, both facing critical choices between incrementally extending existing propositions or fundamentally rebuilding processes and offerings.
- Back-office focus dominates incumbent bank strategies. Despite 99% of surveyed banks prioritizing AI initiatives in customer-facing services, only 32% realize significant returns from these efforts while 68% agree substantial value stems from efficiency gains in back-office operations including KYC processes, compliance workflows, and marketing analytics.
- Fintechs advance faster through digital-native architectures. Building technology stacks from scratch enables fintechs to embed agentic AI throughout value chains from operations and product design to scalable delivery, exemplified by systems like Stripe’s Payments Foundation Model processing $1.4 trillion annually and Arta Finance’s agent-powered portfolio personalization.
Fintechs advance agentic AI adoption more extensively and rapidly through digital-native architectures and cultures of rapid innovation, embedding capabilities throughout entire value chains from operations and product design to scalable delivery. This agility stems from building technology stacks from the ground up. As one neobank executive noted, authoring proprietary core banking platforms means every service follows the same pattern, allowing any engineer to work anywhere in the stack without getting lost.
Notable examples of innovation include:
- Stripe’s Payments Foundation Model processing approximately $1.4 trillion in payments volume annually to detect fraud, optimize payment flows, and predict disputes;
- Arta Finance’s creation of portfolios-of-one for each client, leveraging investment, research, and product agent suites to serve thousands of investors.
Mass Adoption Requires Four Critical Enablers
While current technological advancements laid groundwork for agentic AI in financial services, further progress depends on convergence of four critical enabling factors:
- Infrastructure, including cloud computing platforms and integrated data pipelines must enable seamless scalability and cross-workflow deployment of agentic AI solutions.
- Supportive regulatory environments providing transparent, ethical, and responsible AI guidance and initiatives foster trust and confidence in autonomous system deployment, exemplified by programs like the FCA’s Supercharged Digital Sandbox.
- Digital identity solutions providing secure, scalable identity verification ensure AI agents act on behalf of verified users and interact with legitimate counterparts.
- Trust from businesses and customers in delegating tasks autonomously, supported by ongoing improvements in AI accuracy and explainability, remains essential.
But achieving mainstream adoption will require more than these foundational enablers. It calls for breakthrough use cases creating clear customer value by addressing core pain points or fixing dysfunctional processes. Uber reshaped taxi service markets by proving streamlined ways to hail rides, compelling competitors to follow. Waymo pushes boundaries further by accelerating rollout of autonomous taxi services, continually enhancing learning capabilities.
Financial services require similar breakthrough applications demonstrating clear superiority over existing processes to drive mass adoption.
Three New Economies Emerge from Agentic Transformation
The Oliver Wyman report identifies three transformative customer paradigm shifts driven by agentic AI will reshape customer experiences, operating models, and competitive dynamics:
- The Assistance Economy will see agents delivering entire customer experiences across ranges of engagement models, changing how banks and fintechs interact with customers.
- Adaptive Customer Experiences will enable institutions to deliver interfaces adjusting in real-time to preferences and context, demanding new engagement strategies.
- Agentic Twins will represent owners with agency and trust, becoming key parts of new customer journeys.
Early banking applications illustrate these shifts. Capital One’s Chat Concierge helps customers buy cars seamlessly by profiling preferences—size, brand, performance—suggesting suitable options, estimating prices, optimizing financing, and scheduling appointments with dealerships. Participating dealerships reported up to 55% higher customer engagement.
Adaptive Experiences Create Segments of One
Algorithmic personalization has become baseline expectation, with customers increasingly expecting products and services tailored to unique preferences. However, only 21% of banking customers report being fully satisfied with personalization they currently receive. Market leaders will pioneer next-generation adaptive customer experience interfaces where personalization emerges dynamically, with customer journeys shaped in real-time by agentic AI insights drawing on contextual data such as product search specifics or user behavior.
What it means: As front-end experiences become increasingly dynamic and data-rich, customer-facing interfaces assemble personalized offerings on-the-fly. Traditional distinctions between channel and product blur and potentially disappear.’
Agentic Twins Centralize Identity and Delegated Authority
Today, customer data remains fragmented and replicated across many services and products people use. Despite promises of Assistance Economies and agentic architectures, customers still manually provide data from multiple sources and manage credentials across accounts. Each bank holds customer representations that are incomplete and often outdated, creating inefficiencies and missed opportunities.
What will change: Agentic twins will be owned by individuals as single, protective keepers of aggregated personal information and digital identity, acting only under explicit, predefined permissions granted by owners. Over time, agentic twin authority will extend from routine tasks to complex decision-making—controlling access permissions, selecting products and services, and interfacing with all providers individuals choose. For service providers including banks, this model enables richer, content-driven insights and faster interactions by replacing inefficient data chasing with permissioned collaboration with people’s agentic twins.
Eight Strategic Imperatives for Financial Services Leaders
To secure continued competitive positions in this transformative landscape, banks and fintechs should focus on eight critical actions, according to Oliver Wyman.
The first three actions center on delighting customers, delivering value, and scaling implementations. Organizations must be present in the first new channel since mobile apps by exposing and embedding existing products in new chat-based channels and piloting highly tailored customer journeys, measuring outcomes and rapidly iterating experiences. They should experiment with dynamic experiences by testing adaptive experience rollouts with controlled human validation before iteration, measuring customer outcomes including uptake and product satisfaction while dynamically adjusting.
The next three actions anchor trust foundations. Organizations must embed digital identity by integrating Open Banking and digital identity for secure, transparent data sharing and agent authentication while ensuring alignment with interoperability protocols.
The final two actions build core infrastructure foundations. Evolving data foundations requires establishing robust data pipelines empowering AI agents to anticipate and serve customer needs, addressing siloed and unstructured data while leveraging partnerships and Open Finance to broaden data access. Building effective analytics layers demands creating analytical capability to use customer and clickstream data for delivery of timely, relevant experiences or propositions while implementing feedback loops to maintain insight accuracy and relevance.
