Why isn’t my finance team adopting AI?
It depends on what is actually stopping them. If people tried AI once, got a wrong number and stopped, the problem is training: they were never shown which tasks AI is good at in finance and which it is not. If they are unsure what data they are allowed to use, the problem is policy, and a one-page set of rules fixes more than another demo. If they simply have no time during close, the problem is timing and workload, and you need to give them protected time on real work. Most teams have a mix of all three.
Who this is for
CFOs, controllers and finance managers who bought AI licenses or approved a tool, and now see a handful of people using it while most of the team works the way they did before.
The short version
- Finance teams are trained to distrust numbers they cannot trace, so a single bad answer often ends adoption.
- Unclear data rules make careful people avoid the tool entirely.
- Close and reporting deadlines leave no room to learn something new.
- Generic training does not show finance-specific uses, so people cannot see the payoff.
- Fix the cause you actually have, not the one that is easiest to fix.
Reason 1: One wrong number killed trust
Finance people are paid to be right. When an analyst asks an AI tool to sum a column or pull a figure from a PDF and it returns something wrong, the natural conclusion is that the tool cannot be trusted. That conclusion is half right. AI assistants are unreliable at arithmetic that is not run through a spreadsheet or code, and they can misread tables in scanned documents. They are much better at reading, summarizing, drafting and explaining.
What helps is showing the team the split clearly:
- Good uses: drafting variance commentary, summarizing contracts and invoices, writing memos and emails, explaining a formula, turning a messy note into a clean checklist, preparing questions for budget owners.
- Use with care: extracting figures from documents, always checked against the source.
- Keep in your systems: calculations that go into reports, reconciliations and anything that needs an audit trail.
Once people know where the line is, a wrong answer stops being a reason to quit and becomes a known limit.
Reason 2: Nobody knows what data is allowed
Finance handles payroll, bank details, forecasts and figures that may be material and non-public. Careful people who are not sure whether they may paste a trial balance into a chat window will not do it. That is the right instinct, and it means a vague policy is a blocker.
Write a short, specific rule set: which tool is approved, which account type, which data can go in and which stays out, and who to ask when unsure. As of November 2026, the business and enterprise plans of the major AI assistants state that customer data is not used to train their models by default, but confirm this in your vendor’s current terms and tell the team in plain words. See how to write an AI acceptable use policy and how to keep client data safe with AI.
Reason 3: There is never a good time to learn
Finance runs on a calendar. During close, quarter end and audit, nobody will try a new way of working, and between those periods people catch up. If training happened once and nobody had time to practice, it fades within weeks.
Practical fixes:
- Schedule training in the quiet part of the month, not the week before close.
- Train on last month’s real work, so practice produces something useful.
- Pick two or three tasks for the next cycle and ask each person to use AI on them, with a reviewer.
- Hold a short weekly check-in for the first month to share what worked.
Our guide on why AI adoption stalls covers the habit side in more depth.
Reason 4: The training was not about finance
A general session on prompting shows how to write a poem or plan a trip. A controller watching that cannot see how it helps with accruals or flux analysis. Finance staff adopt AI when they see it applied to their own tasks, with their own kind of documents, by someone who understands why the numbers must tie out.
Good finance training uses examples such as drafting commentary for a variance the team actually explained last month, summarizing a vendor contract to check payment terms, or writing a clear email to a budget owner asking about an overspend. See AI training for finance teams and role-specific vs general AI training.
Reason 5: Leaders are not using it
If the CFO and controller never mention AI, the team reads that as optional. Leaders do not have to be power users. It is enough to use it visibly on one or two of their own tasks, ask in reviews whether AI helped, and share a good example now and then.
How to find out which reason you have
Ask, do not guess. A ten-minute conversation with each person, or a short anonymous survey, usually shows the pattern. Useful questions:
- Have you tried the AI tool on work in the last two weeks? On what?
- What stopped you from using it more?
- Are you clear on what data you can and cannot use?
- Which part of your month would you most like help with?
Then check usage data from the tool’s admin panel if it has one. Low logins point to time or trust. Logins with short sessions often point to people not knowing what to do with it.
Common mistakes
- Buying more licenses when the existing ones are not being used. See why licenses are not a strategy.
- Mandating usage without showing which tasks to use it for.
- Training during close.
- Treating caution as resistance. Careful finance staff are an asset, so give them rules and checks instead of pressure.
- Measuring logins only. Track time saved and rework on specific tasks instead. See measuring whether AI training worked.
How Ruby helps
- AI training: hands-on sessions built on your team’s own close and reporting work, with follow-up during the live cycle.
- AI strategy and consulting: finding out why adoption stalled, writing clear data rules and choosing first tasks.
- AI engineering: building connected workflows when the real time sink is moving and cleaning data, not writing.
If your finance team has the tools but is not using them, book a call with Ruby.
