Keeping pace with the rapid evolution of the finance and accounting services landscape is genuinely challenging. The way F&A work is delivered is being overhauled, with service providers moving from an FTE-heavy model toward one that is agentic and increasingly autonomous. With this in mind, I recently spoke with the HCLTech team about their launch of an autonomous finance platform in partnership with Google Cloud’s Gemini Enterprise.

From people-centric to platform-led

HCLTech has framed its own trajectory as a move away from a people-centric F&A offering toward platform-led operations, which sees a shift in emphasis away from how many people are deployed on a process to how data can be used to deliver universally orchestrated AI agents. Large language models are used to predict and act, and the ecosystem is governed centrally rather than managed on a process-by-process and team-by-team basis.

The autonomous finance curve

Autonomous finance is not new and has been talked about for many years. What has changed now is the trajectory. For the last 10-12 years, straight-through processing (STP) across F&A has improved arithmetically, with incremental gains while leveraging workflow tools and rule-based automation. In the last few years, that curve has spiked and STP is now growing at a rate that looks closer to geometric than arithmetic.

This is the real shift: the narrative remains the same, but the adoption curve has changed. This change turns a long-running aspiration into an operational reality that CFOs must plan for, even though the journey remains long and many details still need to be resolved.

What does autonomous finance mean for a CFO?

Autonomous finance changes how finance work is performed. In the traditional model, F&A delivery was largely FTE-heavy. Hyperautomation helped teams become more productive, but it did not fundamentally change who, or what, performed the work.

In the agentic and autonomous model, solutions shift more of the work to AI agents and an orchestration layer. HCLTech points to a potential 50%-60% productivity improvement, with agents handling more execution.

The enabling architecture in this model is supported by adaptors such as RPA, APIs, connectors, and LLM gateways, which help agents interact with existing systems and processes.

The case for change relies on a common set of problems. For years, fragmented and siloed F&A has driven revenue leakage, cost inefficiencies, and trapped cash flow. HCLTech positions its agentic autonomous F&A solution as an answer to those problems. According to HCLTech’s metrics, the solution is designed to address revenue leakage, improve cost efficiency by 2-3% and improve cash flow by up to 25%.

Realizing those outcomes requires CFOs, their stakeholders, and service providers to reimagine the operating model rather than adding new bolt-on tools to the old one. The potential outcomes can be substantial, such as productivity improvements of up to ~50%, discount-capture improvements of ~40%, DSO reductions of ~23%, and cash-conversion reductions of up to 36%. On the transactional side, the solution points to STP improvements by up to ~35%, exceptions reduced by up to ~35%, and duplicate payments reduced by up to ~90%, alongside an overall productivity gain of ~50%, working capital improvements and margin uplift.

These are potential outcomes rather than guaranteed results. Even so, in the near term, they point to stronger financial discipline. Over time, they could also signal a broader shift in outcomes and create a competitive advantage.

HCLTech's middle path

The more interesting part of the discussion was not just the metrics but the architectural POV behind them. HCLTech is deliberately positioning its offering between two poles. At one end sits the software-centric, legacy way of doing the work, built around workflows and SaaS. At the other is the temptation to go agentic full throttle, stitching together best-of-breed agentic point solutions.

HCLTech's play sits in between. Rather than replacing one set of tools with another, its autonomous finance solution replaces workflows and SaaS or point solutions with an orchestration layer. That layer sits on top of HCLTech’s agents while also leveraging the native agents of the underlying SaaS platforms. This is a more pragmatic and neutral position and one worth watching.

What this looks like in business operations

Underneath this positioning, the solution is an agentic platform that leverages Google Gemini. It brings domain agents for areas such as AP, AR, and banking, coordinated by a multi-agent orchestration layer. Around these sit low-code and no-code automation for rules and task execution, decision engines, and, importantly, a human-in-the-loop rather than a fully hands-off design.

HCLTech supports the autonomous offering with AgentForge, the build layer, and pairs it with change management, discovery and diagnosis through AI Prism and observability through Agent 360. The autonomous F&A offering encompasses 96 reusable AI agents, exception orchestration, a unified finance data layer, a Gemini-powered interface, autonomous systems and decision intelligence, spanning intake-to-pay, order-to-cash, record-to-report and FP&A.

Conclusion

The HCLTech and Google partnership is an indicator of where F&A delivery is headed: from headcount to agents and human-in-the-loop, from process silos to an orchestration layer, from the aspiration of autonomous finance to a curve that is finally progressing geometrically. The economics discussed and the business use case are compelling, but this requires an operating model change to unlock the potential, which is the harder part for most large organizations.

The differentiator over the next 12 to 24 months is not about who has the most agents, but about who can orchestrate across a messy existing estate, govern the resulting system credibly, and carry finance teams through the change, all the while ensuring the GRC requirements are met. The cost of tokens remains under control. On that, the middle-path stance HCLTech has taken is worth noting, as clients decide how far and how fast they want their F&A function to become autonomous.