I caught up with a CFO I’ve known for years last week. Ten minutes in, he shared his screen and showed me a cash dashboard he’d built. Live data, clean, impressive stuff.
Then the grin.
‘Well. I say I built it. Claude built it.’
We laughed. I was in no position to judge… I’m not a finance person, and I’ve used the same tool to build my own cash flow model. Then we talked for an hour about what that sentence means, because he’d spotted the same thing I’ve been seeing all summer. The tools have stopped being the hard part.
Five conversations this week, five finance leaders at different stages, and underneath every one the same admission, usually offered half as a joke. I know this matters. I just don’t know where to start.
Gartner put a number on that feeling in June. 84% of finance organisations have implemented AI or are planning to, and only 7% report a high impact from it. So nearly everyone is in the water, and very few are feeling any benefit yet.
Two camps are shouting at those CFOs right now… The futurists say don’t hire, build an agent. The recruiters say a machine will never replace what a human can do. Both camps are selling something, and both are missing the point.
Every brochure made the same promise
For decades, finance has been promised the shift from preparer to partner. Every ERP implementation, every BI tool, every shared-service redesign came wrapped in the same pitch. ‘Free your team from month end. Elevate them to insight.’
It never quite happened. The tools shaved hours off tasks, and the shape of the work stayed the same. Teams kept preparing, checking, chasing and reconciling, with the partnering squeezed into whatever was left.
The pattern I see across finance teams, so consistent I now treat it as a rule of thumb, is that roughly two thirds of a team’s week goes on mechanical work. Invoices, journals, reconciliations, first-draft reporting. A pattern I’d stand behind, though it’s an observation from the middle of the market rather than an audited figure. The time left for insight is a sliver. Ten percent if they’re lucky. Most teams are just surviving.
What’s changed is that the machine can now do the mechanical work properly, end to end. It does the task, checks its own output, and flags what it couldn’t validate for a human to rule on. I’ve watched that happen this year, in real teams, on real ledgers, with real auditors to satisfy.
So the promise finance never collected is finally payable. I’ve started calling it the human unlock, because the value was always in the people. It’s been trapped under the preparing.
What actually gets unlocked
Sit with what’s left when the preparing goes.
Judgement. The calls where the data gives you odds rather than answers, and a decision still has to be made by Friday.
Translation. Knowing how your CEO needs the missed forecast packaged before the board sees it. The EQ to understand your audience and what matters to them.
Trust. Standing behind a number in a room where standing behind it costs something.
None of that is new work. Finance leaders have carried it on the side of their desks for years, squeezed around the preparing. The unlock is that, for the first time, there’s room to do it properly.
Which is where I part company with both camps. The futurists are right that agents will swallow the mechanical layer, and wrong to think that shrinks the human question. It makes the human question the only question. The recruiters are right that judgement can’t be automated, and wrong to pretend the job description survives unchanged. The two go hand in hand, and almost nobody is standing in the middle saying so.
This is about reimagining what your finance team is there for after they’ve been liberated from the mechanical month end monotony…
Imagine what could be accomplished if they spend the majority of their time on insight and influence? Imagine what impact that would have on margins, revenue, and EBITDA? Imagine what impact it would have on your exit valuation?
It may sound blue sky to some, but I’ve met CFOs that have already reached this destination.
What this means in practice
Four things I’d put in front of any finance leader this autumn.
- Pause before you backfill. When the next resignation lands, resist the reflex to dust off the old spec. Ask what should be automated before a human inherits the role, and what the remaining human work actually is. Sometimes the answer is still ‘hire’. Let it be a decision, taken with the future version of the job in view.
- Pick one output and reimagine it, rather than speeding it up. The forecast you’ve always built, produced in a tenth of the time, is a saving. A forecast that models a thousand scenarios overnight and tells the board which three matter is a new capability. Aim for the second at least once this year.
- Make trust explicit. Finance can’t tolerate errors the way marketing can. Whatever you automate needs validation at every stage, traceability an auditor would accept, and a named human owning the final layer. Skip this and the failure arrives in October, at audit, with an audience.
- Look at your team through the unlock lens. When the preparing goes, who already has the judgement, the translation, the trust-building in them? Who could grow into it with coaching? Where are the gaps no amount of coaching will close? That’s the honest version of workforce planning for the next three years, and very few teams have done it.
Where this leaves you
Whenever a CFO asks me where to start, I end up back at the same sentence. The tech is the easy part. The constraint is the team.
So instead of ‘which AI tools should we adopt?’, try this…
If the mechanical two thirds of your team’s week disappeared over the next couple of years, would the people you have, in the roles you’ve designed, be the team you’d choose for what’s left?
If the answer’s yes, you’re rarer than you think. If it’s no, I hope this gives you some useful food for thought.
If you’d like to think out loud about what the unlock looks like in your team… where to start, what to leave alone… drop me a note at [email protected].
We’ve built an expert team to help CFOs with their AI-enabled finance transformation (both the people and the machine underneath them), and i’m always happy to compare notes.
Thanks for reading,
Leo Hewett
