Guest Articles

From advising to acting

Most AI tools in finance today analyse information and hand a person the output to act on. Agentic AI is different. It takes actions itself, within defined parameters.

In practice that could mean chasing overdue invoices, posting journals, reconciling accounts, processing invoices, or eventually triggering payments, with people involved mainly for the exceptions. The technology is developing quickly, and high-volume, rules-based finance work will be impacted first.

The new boardroom question

This changes the question leaders should be asking. It is no longer only which tools make the finance team faster. It is which decisions and actions you are willing to let an agent take on the company's behalf.

If an agent sends a client communication or triggers a payment, that is still an action of the company. The board still carries responsibility. The auditor still needs an audit trail. Someone is still accountable when something goes wrong. The most important work therefore happens before deployment, in deciding what an agent may do, where the limits sit, and what still requires human approval.

Someone has to be responsible - an that someone cannot be an AI Agent

Some responsibilities should remain firmly human, even when AI can technically perform the task. Setting the rules the AI operates under. Statutory filings. Investor communications. Audit responses. Strategic forecasting and capital allocation. Any decision involving employees.

Board reporting belongs here too. Producing numbers is one task. Knowing which numbers matter, what to challenge, and when the board needs a difficult conversation is another entirely.

The risks of getting it wrong

If many agents are acting across a function, you must still be able to reconstruct why a decision was made and who authorised the rule behind it.

Incorrect logic or poor data can be applied across thousands of transactions before a person notices. And there is what I call human atrophy, where teams grow so used to reliable outputs that they lose the habit of questioning them. The real risk is not a spectacular failure. It is a system that works well enough that people stop checking, and then gets something important wrong.

A UAE priority

This matters more here as regulatory finance becomes technology-driven. Corporate tax, VAT and the arrival of e-invoicing are exactly the areas where agentic AI can apply established rules consistently at scale.

But automation does not remove the need for interpretation. Once a rule is correctly understood, AI applies it efficiently. If the rule has been misunderstood, the AI executes the wrong logic at scale, consistently and confidently. The more you automate in a regulated environment, the stronger the governance around it must be.

The opportunity for growing businesses

None of this is a reason to hold back. Handled well, agentic AI is an advantage, and smaller businesses may gain the most. They are agile and do not have years of legacy to deal with. They can try, fail and move onto something that works much faster. But the governance has to be right.

That is senior finance judgement, and it is why many growing companies will need CFO-level oversight earlier than they expect, even if they cannot yet justify a full-time hire. Capable technology and experienced judgement are becoming more important.

Author: Zaid Aboobaker is Founder and CEO of CompassPoint Consulting, a fractional CFO and financial advisory firm and a member of the British Chamber of Commerce Dubai.