What a year of AI conversations in financial services taught me about where businesses are really standing.
Over the past year I’ve spent more time than ever in rooms with business owners talking about AI. More than 100 of them, and most run businesses in financial services, where compliance isn’t a box you tick once a year. It’s the job.
It’s been humbling. The tech echo chamber would have you believe everyone is either three agents deep or hopelessly behind. Inside the businesses we actually work with, the picture is far more human, and far more varied.
No two of those conversations started in the same place. But it didn’t take long to realise they were all on the same road.
The roadFour places you might be standing
It isn’t a ladder, and nobody climbs it neatly. Businesses skip stops, double back, and occasionally stand in two places at once. But almost everyone we speak to recognises themselves in one of these.
Curious
Leaders know AI matters, but nobody has really started. Experiments are personal and a little bit private.
You’ll hear“We know we should be doing something.”
Tooled up
Licences bought, rollout announced, adoption assumed. Usage is thin and uneven, and the tools mostly sit there being expensive.
You’ll hear“We’ve given everyone Copilot.”
Sprawling
AI is genuinely in use, which is great, across a growing pile of tools and accounts. Nobody owns the list.
You’ll hear“Honestly, I’m not sure what people are using.”
Operating
AI has an owner, a rhythm and a review. It gets measured and improved, then measured again.
You’ll hear“We meet monthly and change things based on what we learn.”
No summitOperating isn’t a finish line. The tools change too fast for anyone to arrive and put their feet up. We’ve tried.
The ditchesTwo ways off the road
The stops weren’t what surprised me. The ditches were. Most businesses that run into trouble with AI do it in one of two ways, and both feel perfectly sensible at the time.
Paralysed
Afraid to overstep. Still used in pockets.
Without clear guidance on what good looks like, the business hits pause. A policy gets written, a few tools get blocked, box ticked.
Meanwhile AI is still being used, just quietly, on phones and personal accounts, by people who still have work to get done. These are often the businesses that feel safest.
Unchecked
Real work. Real client data. Security still catching up.
Something works, so it keeps going. I’ve started more than one conversation with the basics, only to find the client has already vibe coded three apps, all happily pumping client data through them.
The ambition is great. Nobody has stopped to ask how it’s secured, who owns it, or what it’s touching.
Neither ditch is a character flaw. Both are reached by sensible people making reasonable calls.
The trouble is there aren’t many signs on this road, and not many people telling you when you’ve drifted.
Why we drew itThe right conversation, not the prepared one
The biggest lesson of the year is that the same AI conversation doesn’t work for everyone. Come in too strong with a nervous business and you add to the overwhelm. I’ve finished more than one meeting with a client staring at me slightly bug-eyed. Come in too gently with a business that’s racing ahead and you spend an hour on basics while the real risk sits untouched.
The road gives us a shared language for working out where a business is standing before we open our mouths. On the road, we help with the next step. In a ditch, we help it back out. Either way, it gets the conversation it actually needs, rather than the one I happened to prepare.
We don’t have all the answers. Nobody does. But we know how to keep you on the road.
Follow alongFollow the road
Over the coming weeks we’ll take each stop and each ditch in turn: what it looks like from the inside, what keeps businesses stuck there, and what helps them move.
So, which number are you?
Or, if you’re being honest, which ditch is closer? No judgement whatsoever. We’ve visited every one of them ourselves.