Getting AI-ready is a people problem
Most companies I talk to are getting AI in fragments. A summary button in Outlook here, a chatbot on the website there, someone in finance quietly using their own account for spreadsheets. There is no joined-up picture, because the pieces are spread across a dozen systems that were never designed to talk to each other.
The usual answer is a tools project: pick a platform, roll out licences, run a lunchtime training session. That part matters, and it is the part I find easiest to fix. Wiring AI into a business properly, with the data handled sensibly and a person reviewing what goes out the door, is a well-understood job now.
What decides whether it works is the people.
The floor just rose
The research is consistent on one thing. When people get a capable AI assistant, the least experienced gain the most. In the Harvard and BCG study of consultants, the below-average performers improved by around 43 per cent, the top performers by around 17. In a study of customer support agents, the newest staff gained about a third and the most experienced gained almost nothing.
That is brilliant for the new starter. It is a lot harder for the person who spent twenty years getting good at the thing, because the gap between them and everyone else just got much smaller. Their judgement still matters more than ever: someone has to know when the machine is wrong. But the craft they were proud of, the bit that made them the person everyone asked, can now be produced by anyone in seconds. Sometimes better.
Nobody budgets for how that feels
This is the part almost no AI rollout plans for. We talk about upskilling and change management, and we skip straight past the grief.
Think about someone who has spent a career learning to do one thing well, and then watches it done instantly, in front of them, by something that does not get tired. That is not a training need. It is a loss, and people deal with loss badly when nobody acknowledges it: they resist, they quietly stop using the tools, or they worry about it at three in the morning and say nothing at work.
So my suggestion for any company getting serious about AI is a line in the budget that has nothing to do with software: proper support for your staff. Access to a counsellor or therapist, time to talk it through, and leaders who say out loud that this is hard. It costs very little next to the licences, and it is the difference between a team that adopts AI and a team that is afraid of it.
What getting ready actually looks like
- Start with the people, not the platform. Find out what each team actually spends its week on, and what they would gladly never do again.
- Keep humans on every output. AI drafts, people decide. That is safer, and it keeps the expert’s judgement at the centre instead of the edge.
- Redefine the roles out loud. If the craft is changing, say what the job is now, so nobody has to guess whether they still have one.
- Look after people properly. Including the support above.
If any of this sounds familiar, grab me for a coffee. I would love to chat more about it, but fair warning: you might not shut me up.