At Temenos Community Forum 2026, HCLTech used its breakout session on autonomous banking to make a broader point about the direction of the industry. Hosted by Monu Kurien Mathew, SVP and Head of Business Solutions, Financial Services, and Ashish Arondekar, SVP and Global Head of HCLTech Temenos Business, the session argued that banking is now moving beyond front-end digitization and isolated AI use cases into something more structural: a new operating model built around intelligent orchestration, bounded autonomy and continuous change.

Many banks have digitized the customer front door, but much of the middle and back office still depends on manual effort, swivel-chair operations and people stitching together broken processes. Customers may experience a cleaner interface, yet the bank itself often still runs through fragmented workflows, delayed handoffs and high levels of operational drag.  This model has reached its limit. The next era of banking will not be defined by better channels alone, but by how intelligently banks can sense, decide, orchestrate and act across journeys, controls and change itself.

Banking is being rewired for 2030

Financial services is not simply evolving. It is rewiring itself for a very different competitive environment.

By 2030, the winning banks are likely to be more anticipatory, more adaptive and more machine-speed in how they operate. Customers will increasingly value banks that understand context, anticipate intent and reduce effort. Fraud, compliance and risk controls can no longer remain after-the-fact activities. And product changes, policy shifts, fraud patterns and partner models are all moving too quickly for traditional project cycles to keep up. In that environment, autonomy stops looking futuristic and starts looking necessary.

The session framed autonomous banking as a response to four structural pressures already bearing down on banks:

• Economics
• Complexity
• Rising expectations
• Change velocity

Banks need more throughput without proportionate growth in headcount or manual cost. Humans have become the integration layer across too many journeys, which is expensive, fragile and slow. Customers and operations teams want real-time answers and less back-and-forth. And regulation, fraud and platform releases are all changing faster than legacy delivery models can absorb.

What autonomous banking means

Autonomous banking doesn’t mean unsupervised AI. It does not mean replacing bankers with a model and hoping for the best. The session defined it much more carefully: bounded decisioning, orchestration and execution inside clear controls: “Autonomy without boundaries is chaos. Autonomy with controls is scale.”

Many conversations about AI in banking still collapse very different things together. Rule-based automation, GenAI summarization, copilots and chatbots all have a role, but they are not the same as autonomous banking. Autonomy is a continuum, moving from task automation to AI-assisted execution, then to agentic orchestration, bounded autonomy and, eventually, self-optimizing systems. The key design principles are what make this usable in banking: human-on-exception, policy-bounded decisions, evidence-native outputs, cross-system orchestration and continuous learning loops.

In other words, the machine handles the routine and the human handles the material. That is not a limitation of autonomy. It is what makes autonomy bankable.

Start where the pain is visible

The strongest early use cases are not the flashiest ones. They are the journeys where pain is already visible: document-heavy, exception-driven and high-volume processes slowed down by human beings doing integration work between systems.

The session focused on three examples.

• Mortgages

The biggest drag often sits not in decisioning itself but in the middle of the value stream: document intake, validation, re-keying, reconciliation, packaging and rework. The session outlined an autonomous mortgage approach that uses agent-led workflows to pull structured financial data, classify documents, extract evidence, categorize transactions, run calculations and apply affordability rules before a banker reviews the exceptions and makes the credit judgment. The argument is not that underwriting disappears. It is that underwriting is restored by stripping away the manual labor crowding out judgment.

• Autonomous investigations

Here, the pain looks different, but the pattern is the same. Analysts are overwhelmed by alert volumes, false positives, context gathering and closure-note preparation before they can apply real investigative judgment. Autonomy turns analysts from collectors of context into evaluators of risk. Again, the value is not only speed. It is consistency, evidence quality and better use of scarce expert attention.

• Trade finance

This sits in the same family of problems: multi-document, multi-party, discrepancy-heavy journeys where too much effort goes into collecting, validating and packaging information before a specialist can make a meaningful call.

The point across all three use cases is that autonomy creates value by compressing the administrative load around the real decision-maker. 

Trust is the real bridge from pilot to production

The move from pilot to production depends less on model intelligence alone and more on the control envelope around it.

Guardrails are not there to slow AI down. They are what allow banks to scale it with confidence. The session laid out the elements of what it called trusted autonomy: governed data and identity, human oversight and approval, evidence and auditability, policy and compliance controls, model and agent security, plus continuous monitoring, testing and the ability to stop unsafe behavior instantly.

That leads to a more useful leadership question. The issue is not simply whether the model can do the work. It is whether the institution can trust it, govern it, observe it and recover from it when needed. That is the difference between a demo and a banking capability.

Autonomous journeys create pressure on the core

Autonomous banking is not just a journey story. It is also a platform story.

As autonomous journeys scale, more pressure moves upstream into the core. Products and workflows change more frequently. Agents, models and controls introduce more release dependencies. Regression risk rises as more journeys become interconnected. Banks need evidence that every change is safe, traceable and reversible. That is the moment when business autonomy crashes into platform reality. If the Temenos lifecycle is still dependent on manual impact analysis, broad regression cycles and labor-heavy release preparation, the operating model starts to fight itself.

That is why the session argued that the Temenos upgrade lifecycle must also become more autonomous. The goal is not just faster upgrades but faster certainty: autonomous impact analysis, autonomous test generation and prioritization, autonomous deployment sequencing and rollback readiness, and autonomous evidence generation for human approval and audit. The larger business point is that banks can’t run autonomous journeys on top of a manual change engine forever.

The banks that win will make autonomy real

Autonomous banking is not a chatbot, a feature release or an AI overlay. It is a shift in how banks design journeys, controls and change. The argument is that autonomy becomes real only when intelligence meets discipline, innovation meets controls and business ambition is matched by platform readiness.

That creates a practical roadmap for banks. Start where the pain is visible. Design for trust from day one. Build reusable capability patterns. And do not leave the platform behind.

By 2030, the winners are unlikely to be the banks with the most pilots, the most demos or the most hype. They will be the banks that turned autonomy into a governed, auditable and scalable operating model.