Agentic AI in Financial Services: From Process Automation to Decision-Driven Operations

The Autonomous Finance Operating Model

In my previous blog I already referred to an off-conference session in Brazil where I worked with a mid-sized leasing company from Campinas. Just North of Sao Paulo.

In this blog I share some more details on the amazing outcomes.

Given the audience, we explored multiple value areas in the finance industry. Not just the obvious ones like contact centers, but also KYC, lending, claims and onboarding. All domains with high volume, repetition and friction...

In asset-based finance (where most of my work sits) processes like, processes like credit assessment, invoice verification, pricing and collections, are still the backbone of operations.

For next level AI transformation, these are not isolated processes. Once you look at them through another lens, they are connected decision moments and that’s exactly where agentic flows start to outperform traditional systems.

In our Decision Sprints, we don’t focus on improving these processes. We focus on replacing decision-heavy workflows altogether.

A Decision Sprint is a short cycle where business, data and AI come together — not around use cases, but around real decisions. We start by identifying the key decisions and mapping where AI can create actual impact. That’s where (1) speed increases (2) risk becomes dynamic and (3) margin is actively managed. Not in dashboards but in decisions.

During the session, something clicked with the senior leaders from the business around the table,

There was a shared realization unleashing lots of energy: This was not optimization. This is a different operating model. What happened?

After a short demo, our Autonomous Finance Operating Model we moved quickly: From current state to what can be built today for the future. Not to sketch a future world in Asset-Based Finance but something you can build and bring to the business, NOW!

The Autonomous Operating Model (BR) as presented

We mapped a typical leasing underwriting process (still heavily manual, fragmented across teams, full of interpretation and delay). In less than two hours, that same flow was redesigned into an AI-native decision layer. Initial estimates showed: from hours and days --> to minutes

The outcome of this session was clear:

→ documents were interpreted automatically,

→ risks assessed immediately,

→ deviations visible in real time,

→ pricing supported by context, not just rules....

Conclusion?

It turned out to be a decision-driven business. And perhaps that is the most interesting insight of my weeks in the Americas: it is not technology that is transforming financial institutions, but the way decisions are made. And in that respect - Europe is not leading.