Stop writing an AI strategy. Write down your work first.

A familiar story: the board asks for an AI strategy. Someone gathers vendor demos, runs a few pilots and writes a slide of use cases. Six months on, there are some licences and little has changed.

The missing middle

AI does tasks. Strategies describe ambitions. Between them sits the thing almost nobody has written down: the work itself. What comes in, what's done to it, what goes out, and who decides.

The duller alternative

1. Pick one area, say how an enquiry becomes an invoice.
2. Break it into pieces: input → transformation → output.
3. For each piece: how often, how long, how well, and who answers for it.
4. Only then ask which pieces a person, software or AI should do.

What changes

The question moves from "which AI tool?" to "which piece of work, and how would we know it's better?" You have a measure before you buy anything, and a reason to stop if it doesn't improve.

"We don't have time to map everything"

You don't need to. One area at a time is enough to start, and the first map usually shows where work waits, which is often a bigger win than any tool.

Respond

Add your view.

Academics, practitioners, sceptics: tell us where this is right or wrong. Mark reads every response, and the best are published here with your name and role, with our reply.

Your name Email (never published)

Role and organisation Link to your profile or work (optional)

Why you are commenting: your field or experience, in a line or two Your response You may publish my response with my name and role, on this site and in future editions of the book. Remember my details on this device Send my response

We use your email only to reply. Privacy.

Where this leads

Start with solid ground

Stratapault builds a digital twin of your technology, logins and data from your own systems, read-only, with an operator beside you. The solid ground any AI plan needs.