45 Facts About AI Agents in Finance (2026)

Key Takeaways

87% of North American CFOs expect AI to be extremely or very important to their finance department’s operations in 2026 (Deloitte, January 2026).
74% of organisations across industries expect to be using AI agents at least moderately by 2027 (Deloitte, April 2026).
Only 15% to 25% of CFOs have scaled AI across finance functions (Bain & Company, April 2026).
Finance organisations at the scaling, orchestrating or multi-agent stage of agentic AI report moderate or significant improvement on 76% of finance metrics, against 44% for those with no plans or still planning (KPMG, May 2026).
Only 42% of finance organisations are strongly assurance-ready for AI, able to produce AI audit evidence efficiently (KPMG, May 2026).
33% of North American CFOs say the chief information security officer owns AI governance; 19% name the CFO (Deloitte, July 2026).
48% of finance leaders name the accuracy of AI-generated financial outputs as a top concern (KPMG, May 2026).
57% of German companies with 20 or more employees now use AI, the first time a majority does (Bitkom Research, September 2026).

Key Takeaways
How widely are finance teams using AI agents in 2026?
What results do finance teams report from agentic AI so far?
How much are finance teams spending on AI?
Do finance teams trust AI agents and who is accountable?
What are the biggest risks of AI agents in finance?
What do AI agents mean for finance jobs?
How far along are German and European companies?
Why do AI adoption figures differ between surveys?
What this means for finance teams

How widely are finance teams using AI agents in 2026?

Most CFOs use some AI and many plan agents, but scaled agent use in finance is still a minority; the share depends on who was asked and what counts.

87% of CFOs at North American companies with at least US$1 billion in revenue expect AI to be extremely or very important to their finance department’s operations in 2026, in a survey of 200 CFOs (Deloitte, January 2026).
93% of North American CFOs say their organisations use AI extensively or modestly; this counts any AI use, not agents (Deloitte, July 2026).
54% of North American CFOs say integrating AI agents into finance will be one of their top finance transformation priorities in 2026 (Deloitte, January 2026).
44% of North American CFOs use AI for financial planning and budgeting, and 41% use it to analyse financial data for insights (Deloitte, July 2026).
52% of finance professionals regularly use AI tools in their roles, while adoption of more advanced agentic AI remains limited, in a survey of 11,389 accountancy and finance professionals in 160 countries (ACCA, May 2026).
74% of organisations expect to be using AI agents at least moderately by 2027, in a survey of 3,235 IT and business leaders in 24 countries across all industries (Deloitte, April 2026).
Half of the 163 US finance leaders in KPMG’s survey already plan to orchestrate or develop multi-agent AI systems across their workflows, and 93% expect to deploy or scale AI in finance within 18 months (KPMG, May 2026).
Only 15% to 25% of CFOs have scaled AI across finance functions, in a survey of more than 100 CFOs worldwide (Bain & Company, April 2026).
52% of financial services industry respondents have agentic AI in active adoption, a figure that includes pilots: 57% of fintechs and 45% of incumbents (Cambridge Centre for Alternative Finance, April 2026).

‍

What results do finance teams report from agentic AI so far?

Organisations further along with agents report clearly better results than those not yet deploying, though the figures are self-reported and satisfaction is still mixed.

‍

Agent deployers report better results

No plans44%

Agent deployers76%

Source: KPMG, May 2026, self-reported

‍

76% of finance metrics show moderate or significant improvement at organisations deploying agentic AI in finance (scaling, orchestrating or multi-agent stages, 537 respondents), against 44% at organisations with no plans or still planning (146 respondents): a 32-point gap, self-reported in a survey of 1,013 senior finance leaders in 20 countries (KPMG, May 2026).
Organisations at the orchestrating and multi-agent stages report close to 40 points more improvement than those still in early planning on forecast accuracy and ROI, which KPMG calls the metrics that matter most (self-reported) (KPMG, May 2026).
36% of organisations that made progress with agentic AI in the past six months name capacity for growth as the most cited source of measurable value, ahead of customer experience at 35% and greater responsiveness at 32% (base of 867) (KPMG, May 2026).
Nearly three-quarters of finance organisations report AI ROI meeting (46%) or exceeding (28%) their expectations (KPMG, May 2026).
31% of CFOs are satisfied with their AI outcomes; over 40% are highly satisfied among CFOs deploying some form of AI at scale, against 25% at companies still piloting AI (Bain & Company, April 2026).
7.5 fewer days to complete the monthly close are linked to generative AI use, not AI agents, in a 2025 field study based on transactions from 79 small and mid-sized businesses and a survey of 277 accountants (MIT Sloan and Stanford GSB via CFO Dive, August 2025).

The gap between deployers and non-deployers is wide, but almost every result here is self-reported. Only the 2025 close study measured time directly, and it looked at generative AI.

How much are finance teams spending on AI?

Budgets are rising, and in Germany the major cost items are infrastructure, data preparation and integration into existing systems.

83% of CFOs plan to increase enterprise-wide AI spending by more than 15% over the next two years (Bain & Company, April 2026).
42% of CFOs expect to raise AI budgets by 30% or more over two years (Bain & Company, April 2026).
45% of German companies with 20 or more employees spend more on AI in 2026 than the year before, 13% significantly more, while 50% keep spending unchanged (Bitkom Research, September 2026).
51% of German companies that use AI name infrastructure such as cloud and computing power as a major AI cost item, 50% name data preparation and 41% integration into existing systems; the base is AI-using companies only (Bitkom Research, September 2026).
More than half of nearly 750 US corporate executives, mostly CFOs, say their firms have already invested in AI; the largest labour productivity effects are in high-skill services and finance, and perceived gains exceed measured gains (Federal Reserve Bank of Atlanta and Duke Fuqua, March 2026).

Do finance teams trust AI agents and who is accountable?

Governance trails deployment: few organisations have mature controls, ownership is split, and human oversight is being increased.

‍

Who owns AI governance?

Information security officer33%

CFO19%

CEO12%

Source: Deloitte, July 2026

‍

59% of North American CFOs say balancing pressure to deploy AI quickly with managing its risks is their biggest AI governance challenge (Deloitte, July 2026).
33% of North American CFOs say their chief information security officer owns AI governance; 19% say the CFO holds the greatest responsibility and 12% the CEO (Deloitte, July 2026).
Only 21% of organisations say they have a mature governance model in place for agentic AI, across industries (Deloitte, April 2026).
42% of finance organisations are strongly assurance-ready for AI, meaning they can produce AI audit evidence efficiently; among agentic AI leaders the share is 60% (KPMG, May 2026).
Only 29% of finance organisations formally track AI adoption failures (KPMG, May 2026).
94% of organisations already rely on third-party assurance providers for AI (KPMG, May 2026).
33% of organisations are increasing human-in-the-loop oversight as a direct response to AI-related concerns (KPMG, May 2026).
48% of the 130 financial regulators surveyed say they are still in the “Exploring” stage of AI adoption or not engaged with AI at all (Cambridge Centre for Alternative Finance, April 2026).

Controls are the gap: fewer than half can evidence their AI for an auditor, so ask how an agent’s work is approved and logged before it touches the ledger.

What are the biggest risks of AI agents in finance?

Accuracy of outputs, security and unclear costs top finance leaders’ concerns, and a 2025 analyst forecast expects many agentic AI projects to be cancelled by 2027.

48% of finance leaders cite the accuracy of AI-generated financial outputs as a top concern, and 50% cite cyber and AI-native security threats (KPMG, May 2026).
25% of North American CFOs are concerned about the potential for AI-driven mistakes (Deloitte, July 2026).
46% of North American CFOs say cost uncertainty or lack of transparency is their biggest internal AI concern (Deloitte, July 2026).
72% of accounting professionals in a year-long study at one mid-tier UK firm worry that generative AI could produce errors or incorrect decisions (ICAS, March 2026).
Over 40% of agentic AI projects will be cancelled by the end of 2027, Gartner predicted in June 2025; due to escalating costs, unclear business value or inadequate risk controls; this is a 2025 forecast across industries, not a measured failure rate (Gartner, June 2025).

What do AI agents mean for finance jobs?

Staff use AI regularly and feel able to learn it, but many worry about their jobs; the evidence on actual headcount effects is still thin.

74% of accounting professionals in the same ICAS study at one UK firm say generative AI speeds up their tasks, but only 22% report a substantial rise in their overall work rate (ICAS, March 2026).
Only 21% of accounting professionals in the same study are concerned that generative AI might affect their job security (ICAS, March 2026).
51% of finance professionals remain worried about the potential impact of AI on their jobs, with younger generations most concerned (ACCA, May 2026).
43% of finance professionals say their employer provides AI-related upskilling, up from 32% in 2025 (ACCA, May 2026).
82% of finance professionals feel confident in their ability to learn and apply AI capabilities (ACCA, May 2026).
45% of German companies expect AI to reduce their headcount slightly over the next two years, and 12% expect a significant reduction (Bitkom Research, September 2026).
473 US firms answered the CFO Survey in February and March 2026, and the researchers find little evidence of substantial AI-driven employment decline, although larger firms anticipate a shift away from routine clerical jobs (Federal Reserve Bank of Richmond, March 2026).

Expectations and evidence diverge: many German firms expect fewer staff, while US survey data shows little decline so far.

How far along are German and European companies?

Over half of German companies now use AI, as do over half of large EU enterprises, yet AI agents are still at an early stage in Germany.

57% of German companies with 20 or more employees use AI, the first time a majority does, up from 36% a year earlier and 20% two years earlier, in a survey of 603 companies (Bitkom Research, September 2026).
54.5% of German companies use AI in their business processes, up from 40.9% a year earlier, and 67.2% of large companies do (ifo Institut, June 2026).
Only 4% of German companies say AI is not a topic for them, down from 17% a year earlier and 41% two years earlier (Bitkom Research, September 2026).
19.95% of EU enterprises with 10 or more employees used AI technologies in 2025, and 55.03% of large enterprises with 250 or more employees did (Eurostat, December 2025).
11% of German companies that use, plan or discuss AI already use AI agents, 29% plan to and 31% are discussing them; the base is not all companies, only those with AI on the agenda (Bitkom Research, September 2026).

AI use in Germany has passed the halfway mark; agent use is still a small group, with a larger group planning.

Why do AI adoption figures differ between surveys?

Surveys ask different people different questions: company size cut-offs, survey year, industry and the definition of adoption all change the result, so compare bases before comparing numbers.

Definitions differ: how many German companies use AI. ifo (June 2026) finds 54.5% of German companies use AI in business processes, Bitkom (20 or more employees, September 2026) 57%, and Eurostat (10 or more employees, 2025 data) 19.95% for the EU as a whole. Size cut-offs, survey year and wording differ between these surveys.

Definitions differ: how many companies run AI agents. Bitkom’s 11% counts German companies using agents among those that use, plan or discuss AI. Bain’s 15% to 25% counts CFOs who have scaled AI across finance functions. Cambridge’s 52% counts active adoption in financial services, including piloting. KPMG compares deployers with organisations that have no plans or are still planning. The bases differ, so do not compare the numbers directly.

Definitions differ: AI use versus AI agents. Deloitte’s 93% includes modest use of any AI. ACCA’s 52% is regular use of AI tools by individual professionals. Neither measures agents that act on their own.

Methodology: how we chose these facts. Facts come from primary research by consultancies, Big Four firms, universities, central banks, statistics offices and professional bodies. 2026 sources come first; a 2025 source appears only where no 2026 figure exists: the Gartner forecast (June 2025), the MIT Sloan and Stanford GSB close-time study (reported August 2025) and Eurostat’s 2025 enterprise data (published December 2025). Surveys commissioned by software vendors are not used. Each fact links to the original publication, except the close-time study, which links to CFO Dive’s report. Survey figures are self-reported unless stated. Facts checked 28 Oct 2026; this page is updated quarterly. By the Transformance research team.

What this means for finance teams

The facts point to one test for any finance AI agent: it should show its evidence, wait for a person’s approval and leave an audit trail an auditor can follow.

Adoption is broad and governance is thin, so the controls around an agent matter as much as the work it automates.

In order-to-cash, solutions like Transformance are built around that test. Vero, the AI agent, works on Customer 360, one graph per customer that serves as its working memory. It investigates an exception such as an unmatched payment, fetches information from internal and external sources, and proposes a resolution with the evidence; a person approves.

Cash application: a 95% auto-match rate, with 100% posting validation; nothing posts without human sign-off (Transformance customer data).
AR collections: DSO down 8-12 days on average (Transformance customer data).
Calls: an AI calling agent that makes 15-20 calls an hour in 70+ languages; a human collector makes 15-20 a day (Transformance platform data).
Documents: over 97% extraction accuracy across 100+ document types, with no template setup (Transformance platform data).

The controls sit around every step. A 4-level approval model (read, recommend, execute, post) keeps a human in the loop for ERP postings, with thresholds by amount, entity and customer, and roles that separate who configures rules from who approves. The audit log records the timestamp, what the system did, what the user did, who approved, the input and the confidence. Data stays in your region, and no customer data is used to train models.

Who Transformance is for

ERPs: SAP S/4HANA, SAP ECC, NetSuite, Microsoft Dynamics 365 Finance & Operations and Business Central, and Odoo in production with customers; Oracle and Infor supported.
Size: mid-market to large enterprise, 250 to 2,000+ employees, including groups with several entities and shared service centres.
Payment volume: from under 1,000 to over 100,000 payments a month.
Industries: manufacturing, consumer goods and food and beverage, logistics, e-commerce, agriculture, energy and professional services.
Scope: cash application alone, or the full order-to-cash (cash application, deductions, AR collections, cash forecasting).

Conclusion: adoption is running ahead of control

AI agents in finance are moving into production faster than governance is maturing: 15% to 25% of CFOs have scaled AI, and only 42% of finance organisations are strongly assurance-ready.

Teams that close the gap build approvals and audit evidence into the rollout from the first day. Book a Call to see how an agent with human sign-off works in your order-to-cash.

Key terms

Agentic AI: AI that plans and carries out multi-step tasks toward a goal, within the permissions it is given.
AI governance: the policies, roles and controls that decide how AI is used, checked and held to account.
AI hallucination: an AI output that sounds plausible but is not supported by the underlying data.
Audit trail: the record of who or what did which step, when, and who approved it.
Explainable AI: AI whose outputs come with reasons and evidence a person can check.

Related guides

‍

Frequently Asked Questions

What percentage of finance teams use AI agents?

It depends on the base: 15% to 25% of CFOs have scaled AI across finance functions (Bain & Company, April 2026), and 11% of German companies with AI on the agenda already use agents (Bitkom Research, September 2026). 74% of organisations expect to use agents at least moderately by 2027 (Deloitte, April 2026).

Are AI agents safe to use in finance?

Safety depends on the controls around the agent, and most are still being built: only 21% of organisations have a mature governance model for agentic AI (Deloitte, April 2026), and 42% of finance organisations are strongly assurance-ready (KPMG, May 2026).

Which finance tasks do CFOs use AI for?

Planning and analysis lead: 44% of North American CFOs use AI for financial planning and budgeting, and 41% to analyse financial data for insights (Deloitte, July 2026).

Will AI agents replace finance jobs?

The evidence so far shows little substantial AI-driven employment decline in the US (Federal Reserve Bank of Richmond, March 2026), while 45% of German companies expect a slight and 12% a significant headcount decrease over two years (Bitkom Research, September 2026).

What results do companies see from agentic AI in finance?

Organisations deploying agentic AI report improvement on 76% of finance metrics, against 44% for those with no plans or still planning (KPMG, May 2026), and nearly three-quarters of finance organisations say AI ROI meets or exceeds expectations (KPMG, May 2026). Both figures are self-reported.

Who is responsible for AI governance in finance?

Ownership is split: 33% of North American CFOs say the chief information security officer owns AI governance, 19% the CFO and 12% the CEO (Deloitte, July 2026).

How many German companies use AI agents?

11% of German companies that use, plan or discuss AI already use AI agents, 29% plan to and 31% are discussing them (Bitkom Research, September 2026). AI in general is used by 57% of German companies with 20 or more employees (Bitkom Research, September 2026).

SAP, SAP S/4HANA, Microsoft, Dynamics 365, Business Central, Oracle, NetSuite, Odoo and Infor are trademarks of their respective owners (SAP SE, Microsoft Corporation, Oracle Corporation, Odoo S.A., Infor). Transformance is not affiliated with or endorsed by them.