AI agents must be held to the same standards as finance teams

Finance teams adopting AI agents must be able to demonstrate how those agents perform accounting tasks, what controls govern their work and who is responsible when something goes wrong.

That was a central theme of the Audit-Ready Finance in the Age of AI panel at TakeControl London 2026, which examined how established internal-control principles can be applied to AI.

Or to put it another way, the technology may be new, but the underlying responsibilities remain the same. Which means someone still has to take responsibility when an AI agent makes a material error.

‍Accountability must stay with the finance team

Of course, this isn’t new. When I was a finance director, there was no way that I could have told my CFO that an incorrect journal entry wasn't my responsibility just because somebody else on the team had posted it.

The same applies to AI. But rather than pin it on someone in their team, finance leaders need to understand the processes their agents perform, provide the right context and ensure that appropriate reviews take place.

This is where the expertise of Lucia Wind, FloQast's SVP of Risk & Audit Advisory, provided valuable insight during the panel. Lucia also chairs the Committee of Sponsoring Organizations of the Treadway Commission (COSO)*, which develops widely used frameworks for internal control and risk management. 

She explained how COSO's guidance helps organisations assess the risks of AI-driven accounting processes and put appropriate controls in place. 

Crucially, she explained how FloQast is applying that approach within our own Risk and Compliance platform, enabling finance teams to assess AI agents, develop risk-control matrices and capture evidence of their work.

Although regulatory requirements differ between the UK, Europe and the US, finance teams still need to assess whether errors could materially affect their financial statements, whether controls can be overridden and whether duties are adequately segregated.

Reaping the rewards of getting it right 

If I’m honest, I could happily write pages about regulation and compliance because it’s a subject that is fundamental to the work we do in finance. But while some people came to TakeControl London 2026 to explore those issues, I’m also conscious that others wanted to hear what others were doing and what results they were achieving. And there were some impressive examples.

For instance, Japanese pharmaceutical company Shionogi has reduced its EMEA financial close from four days – to just one.

The improvement came from rethinking existing processes, establishing clearer workflows and giving the finance team better visibility of progress. The result is that the team can now celebrate completing its close at 5 pm on day one rather than working late to meet deadlines.

PageGroup improves visibility and accountability

Elsewhere, the finance team at PageGroup – the global recruitment group behind brands including Michael Page – introduced reconciliations, checklists and variance analysis to bring greater consistency to its processes.

As a result, recorded sign-offs give managers clearer visibility of completed work and who is responsible. The company is also moving towards greater automation through AI-powered matching.

An important lesson from their experience was the need to involve finance staff in changes to their processes, rather than simply introducing new technology and expecting people to adopt it.

Group 1 Automotive builds confidence through standardisation

And finally, international automotive retailer Group 1 Automotive faced the challenge of standardising finance processes across different brands.

Rather than switching immediately to a new way of working, the company ran its existing processes alongside the new ones for around three months. That gave its teams time to check the results and build confidence in the changes.

The three examples show that improving finance operations is not simply a question of adopting new technology. Clear processes, visibility and accountability are what allow teams to work more efficiently and trust the results.

Establishing the right controls is critical to success

What struck me about these discussions was how much they had in common. Whether you're introducing AI agents or trying to close the books faster, you need to understand the process, establish the right controls and make sure people know what they're responsible for.

And if you're bringing AI into the equation, auditability needs to be built in from the start, not considered once the technology is up and running.

Get those fundamentals right, and you give your finance team the confidence to work differently, whether that means adopting AI or getting everyone home on time at the end of the close.