For the seventh year, we have asked finance leaders and professionals where they really stand with AI and automation.
This year’s report is called The Age of Exploration because finance is no longer standing at the edge of AI, wondering whether to begin. Most organizations have already moved into the territory. They are investing, piloting, deploying, and, in some cases, allowing AI to execute real work. What they are discovering now is that the difficult part begins once the technology has to operate inside real finance processes.
Our 2026 State of AI Automation in the Finance Office report found that 66.5% of finance organizations are increasing their investment in AI, while only 0.7% are reducing it. Almost a quarter (24.2%) now consider AI a top budget priority. Yet only 21.0% report meaningful, measurable success, while 64.8% describe their results as mixed or unsuccessful.
That gap is one of the most important findings in this year’s research.
What stood out this year
A few findings tell the story particularly well:
- 58.4% of finance teams are still exploring or piloting AI, while only 12.8% have reached the optimizing or autonomous stages.
- AI use is widening quickly. The average finance team now uses AI across 2.43 functions, a 36% increase in one year.
- Finance is replying faster but spending far more time chasing people. Some 43.8% of requests are answered within 24 hours, yet 44.5% of teams now spend 11 or more hours a week following up with vendors, customers, and internal stakeholders.
- AI is beginning to execute work. Some 39.9% of finance teams now use an operating model where AI carries out work, either with human review of exceptions or within defined rules.
- The barriers are changing. Integration with existing systems now leads at 36.7%, followed by high initial investment costs at 36.3%, security and privacy at 34.5%, finding suitable automation tools at 33.8%, data quality at 31.3%, and AI governance and transparency at 30.2%.
We need to measure impact, not activity
One of the easiest traps with AI is to confuse movement with progress. Finance teams can count pilots, use cases, tools deployed, and agents introduced. Those numbers show activity. They do not show whether the business is getting better outcomes.
For us, that means looking at whether AI is helping finance complete work faster, make better decisions, reduce unnecessary intervention, and move cash through the business more effectively. That is particularly important given how much broader AI deployment has become. Finance teams are now using AI across areas, including document digitization, anomaly detection, cash flow forecasting, compliance reporting, and customer and vendor inquiries.
Among organizations significantly increasing AI investment, 60.3% report successful initiatives. Among those increasing investment only modestly, that falls to 10.1%. The research does not prove which causes which, but it does show that deeper commitment and stronger results appear together.
Moving AI into day-to-day finance operations also requires work beyond the budget itself, including integration, process redesign, and governance.
Faster tasks are not enough
One of the clearest findings in the report is the disconnect between response speed and the amount of effort still required to get work completed. Finance teams are answering more quickly than they have in years.
Yet the amount of time spent following up has risen sharply, with 44.5% now spending 11 or more hours a week chasing vendors, customers, and colleagues. Another 13.4% spend more than 30 hours a week doing it. That tells us the problem is not simply speed.
A document can move faster. An AI system can draft a response. A workflow can send an approval request instantly. But the work can still stop at the next handoff because information is incomplete, another approval is needed, or the next system does not have the context required to continue.
This is particularly visible across the Cash Cycle, where a single invoice or payment can touch multiple systems, people, approvals, and exceptions before the work is complete.
Trust changes when AI can act
The conversation also changes once AI moves from providing an answer to taking an action. Our research found that 26.3% of organizations are already using supervised autonomy, where AI handles routine work, and people step in for exceptions. Another 13.5% describe themselves as operating with governed autonomy, where AI can execute work end-to-end within defined rules and compliance guardrails.
That requires a different level of trust. Finance needs to know what authority AI has, which systems and information it can access, which policies apply, when it must escalate, and what record remains after the action is taken.
That is what Auditoria means by Governed Autonomy. The goal is not to remove people from finance. It is to give AI the authority it needs to complete appropriate work while keeping that authority within the controls and accountability that finance already expects.
Finance is starting to rethink how the function works
Perhaps the most encouraging change in this year’s research is that finance leaders appear to be looking beyond the technology itself. Process improvement remains the top priority, while evaluating functional strategy, scope, and design rose sharply to 19.9%, the largest movement on the priority list this year.
Adding AI to an old process does not automatically improve it. If the process is still fragmented, approval-heavy, and dependent on manual handoffs, AI simply speeds up pieces of the same workflow.
Finance now has to rethink how that work should move, what AI should be trusted to handle, where people still need to step in, and how the function itself should change as more routine work is automated.
The map is no longer blank. Finance can see more clearly where AI is working, where it is getting stuck, and what still has to change. The next part of the journey is about making AI work across the Cash Cycle, inside the systems, controls, and processes that finance already depends on, and being able to show that it is improving how the function actually operates.
The full 2026 State of AI Automation in the Finance Office Report, The Age of Exploration, is based on 292 validated respondents, with a base of 281 for subsequent survey questions.
Frequently Asked Questions
What is the overall state of AI adoption in finance in 2026?
Most finance organizations have moved past the question of whether to adopt AI and are now actively investing, piloting, and deploying it. However, only 21.0% report meaningful, measurable success, while 64.8% describe their results as mixed or unsuccessful.
How widely are finance teams investing in AI right now?
66.5% of finance organizations are increasing their AI investment, and nearly a quarter consider it a top budget priority. Only 0.7% are reducing their investment.
Why are so many AI initiatives in finance producing mixed results?
The leading barriers include integration with existing systems, high initial investment costs, security and privacy concerns, and data quality challenges. Adding AI to fragmented, approval-heavy processes does not automatically improve them.
What does “Governed Autonomy” mean in the context of finance AI?
Governed Autonomy means giving AI the authority to complete appropriate work end-to-end while keeping that authority within the controls, compliance guardrails, and accountability that finance already expects. It is not about removing people from finance, but about defining clear boundaries for what AI can act on independently.
How is AI use expanding across finance functions?
The average finance team now uses AI across 2.43 functions, a 36% increase in just one year. Use cases include document digitization, anomaly detection, cash flow forecasting, compliance reporting, and vendor and customer inquiries.
Why are finance teams spending more time on follow-up even as response speeds improve?
43.8% of requests are answered within 24 hours, yet 44.5% of teams now spend 11 or more hours a week chasing vendors, customers, and colleagues. Work still stalls at handoffs when information is incomplete, approvals are missing, or the next system lacks the context to continue.
How many finance organizations have AI executing real work today?
39.9% of finance teams now operate a model where AI carries out work, either with human review of exceptions or within defined rules. Only 12.8% have reached the optimizing or autonomous stages of deployment.
What should finance leaders prioritize to move beyond mixed results?
Finance leaders need to look beyond the technology itself and rethink how work should move, what AI should be trusted to handle, and how the function should change as more routine work is automated. Process improvement remains the top priority, and evaluating functional strategy and design rose sharply this year.