Last week, Infosys Finacle brought clients, technology partners, and analysts to Rome for its Finacle Conclave. Few cities illustrate the challenge of modernization better than Rome. As the city expands its metro network, every attempt to build for the future must contend with layers of infrastructure and history already beneath its streets. Banks face their own version of that challenge, making Rome a fitting setting for Finacle Conclave 2026 and its focus on how banks can compose an AI-native future from the complex technology foundations they already have.

The event’s theme, “Composing AI-Native Banking,” captured the shift. While AI dominated the conversation, the bigger question was what banks themselves need to become when intelligence is embedded throughout the enterprise. As we wrote in our recently published Predictions 2027: Banking And Investing, banks can’t achieve AI-native banking simply by layering AI onto legacy technology, fragmented data, and existing processes. They need to rethink the architecture, operating model, processes, and ecosystems that underpin banking.

AI-Native Banking Requires More Than AI

The most important takeaway was perhaps the simplest: AI-native banking isn’t primarily about AI — it depends on the capabilities around it.

Infosys Finacle’s vision combines real-time data, AI platforms, domain-specific intelligence, agent orchestration, APIs, events, model context protocol (MCP)-based tools, cloud infrastructure, and governance. Its model moves from traditional banking through digital banking toward composable AI-native banking — where humans and AI collaborate, propositions become more dynamic, and banks participate in ecosystems involving fintechs, platforms, and AI agents. The scale of AI adoption discussed at the event shows why those foundations matter: One large European bank has already deployed more than 150 AI use cases across its business, while another bank reported 63 AI initiatives in production and more than 150 processes being redesigned around AI.

That framing matters. Banks have spent years digitizing channels and individual journeys. AI raises the stakes because intelligence increasingly needs to operate across journeys, products, channels, and systems. Fragmented data, tightly coupled applications, disconnected capabilities, and inconsistent customer context therefore become more than technology shortcomings: They constrain what AI can actually do.

One large Middle Eastern bank illustrates why becoming AI-native requires more than deploying AI. The bank highlighted the difficulty of creating a holistic approach to personalization when marketing, sales, and service functions invest around their own objectives, creating fragmented technology and customer experiences. This is precisely the organizational problem AI-native banking needs to overcome: Intelligence needs shared customer context and capabilities that work across functional boundaries, not another layer of point solutions.

Composability Is Becoming A Prerequisite For AI

Infosys Finacle presented a banking-focused AI stack extending from infrastructure and data through models, applications, agent orchestration, and experiences. Banking capabilities can be exposed through APIs, events, and MCP servers so they can be reused and orchestrated in different contexts. For example, it described a Skills Hub containing more than 200 governed banking skills that can be called as MCP tools and orchestrated into workflows.

This points to a broader shift. Banks need to move from designing around events and channels toward composing capabilities around customer intent. Rather than building another vertically integrated journey each time a customer need changes, banks need reusable capabilities that can be combined across their own experiences and, increasingly, external ecosystems.

One large Latin American banking group offered a tangible example. Its approach separates the commercial proposition customers encounter from the underlying financial product and moves from thinking about a single core toward a stack of platforms spanning customers, processes, risk, products, payments, finance, data, and AI.

Core Modernization Is Becoming An AI Issue

For years, banks have justified core modernization through agility, cost, resilience, and speed to market. AI adds another reason.

The discussions repeatedly connected legacy modernization with banks’ ability to exploit AI. Legacy estates contain valuable business logic and institutional knowledge, but tightly coupled architectures also make it harder to expose that knowledge, connect real-time data, and compose capabilities dynamically. The implication isn’t that every bank needs a wholesale core replacement.

One Australian bank illustrated the trade-off particularly well. Instead of reproducing two decades of legacy practices in its new environment, the bank chose to standardize around an out-of-the-box implementation, accepting greater organizational change upfront in exchange for less structural complexity over time. It moved from project approval to go-live in just 15 months, redesigning more than 60 procedures and migrating more than 14 million data records in the process.

The lesson for banks is important: Don’t use modernization to recreate yesterday’s bank on newer technology. Build the foundations for what the bank needs to do next.

AI Should Redesign Processes, Not Automate Their Complexity

Another message resonated strongly: Don’t automate complexity — remove it.

As banks move from ML to generative and agentic AI, the opportunity is increasingly to rethink entire processes rather than make individual tasks marginally faster. One bank described how agentic workflows now complete 12 of the 18 steps in its corporate KYC process, illustrating the shift from using AI to assist individual tasks to redesigning end-to-end workflows.

This changes how banks should measure AI progress. The question shouldn’t merely be how many AI use cases they deploy. It should be whether AI helps them remove handoffs, simplify processes, make better decisions, personalize propositions, and deliver better customer and business outcomes.

Banking Leaders Need To Modernize For AI, Not Just Adopt It

Deploying more AI won’t make a bank AI-native. As our research on high AI adopters shows, the organizations making the most progress aren’t necessarily those deploying the most pilots. They distinguish themselves through customer-led AI strategies, committed leadership, and readiness across data and platforms. Banking leaders need to tackle the architecture, data fragmentation, process complexity, and organizational silos that limit what AI can accomplish. Treat AI as a catalyst to simplify the bank: Modernize the foundations, turn business capabilities into reusable building blocks, redesign processes around outcomes, and establish clear boundaries for AI and human decision-making.

Forrester clients can schedule a guidance session with us to discuss what AI-native banking means for their modernization strategy and priorities.

Share