Start with one real job.
Pick a task with a clear owner, clear inputs and a result you can see.
Find the right job. Prove the value.
Build from what pays back.
You've been told you need AI. You may already have a CRM, a portal and a dashboard nobody opens, so the scepticism is earned. My approach is simpler: start with one real job, prove the value, and let it earn the next.
Reformat Labs is my AI consultancy for UK businesses. What I learn there, I write about here: what's worth trying, what to leave alone, and what it actually takes to make one job stick.
Read about my approachPick a task with a clear owner, clear inputs and a result you can see.
Agree what good looks like, test it with the people doing the job, and fix what breaks.
Set success before you start. A proven result gives you the confidence and evidence to take on the next job.
The best first project is simple enough to launch, important enough to matter, and visible enough to prove.
AI adoption is a business habit, not an IT handover. The people doing the job must shape and improve it.
Give saved hours back to customers, quality and work that has been waiting, not just to a cost-cutting target.
Working together
A plan for the first job, the number it has to hit, and what comes after it. Done with the person who can say yes in the room.
Start a conversationA working session with your team. You leave with one job chosen, one measure agreed and a first step you can take next week.
Plan a workshopTalks and panels for business audiences who want to know what AI actually does inside a working business, and what starting involves.
Enquire about speakingThree people have to agree the standard before an AI step touches real work, and it has to be a standard a reviewer can apply in seconds.
The four lines an AI tool costs you every month, and the short review that keeps all of them in view.
How far a wrong answer travelled decides what it costs you, and what you should do in the hour after you spot it.
If you have a job in mind, a question about AI, or a business problem worth working through, tell me what you’re trying to make work. I’ll help you find the sensible next step.
Start with the job
A short conversation is often enough to work out whether AI can earn its keep.
jay@jayjanes.com