Your AI strategy doesn’t need a capital S

It’s planning season, and the companies I talk to are planning next year’s digital transformation and AI adoption. For many of them that means a strategy first: months of meetings, a committee, often a big consulting firm, and only after that the first real project.

It doesn’t have to be Strategy with a capital S, and it doesn’t have to be handed to a big, expensive firm. You know your business better than anyone you could hire. What you need is a good process, and you can run it yourselves. We wrote the whole process up as a free kit you can use this week: the questions to ask, a template, and AI prompts for each step.

What a strategy actually is

A strategy isn’t a document. To a large extent it is the sum of the things your company is working on, all pointed in the same direction. Richard Rumelt, in Good Strategy Bad Strategy, puts its core in three parts:

1. A diagnosis: what is actually wrong, or standing in the way.
2. A guiding policy: the overall approach to dealing with it, which rules some options out.
3. Coherent action: steps that support each other, with everyone pulling in the same direction.

Saying it is the easy part. Each part has its own way of going wrong: a diagnosis that is really a wish list of solutions, a policy that is just a target (“grow 20%”), and actions that pull in different directions, or start with a project too big to finish. A good process guards against all three. The week below does the diagnosis, lines it up with leadership’s goals, and ends with the first action.

Waiting for the big version has a cost too. Every month spent planning is a month the same problems keep costing you. And the plan gets better once something is running. When people see a process that has always worked one way suddenly work ten times better, they start seeing what else could change.

The process: one week

Before Paloma, I worked at a high-growth tech company where we planned the roadmap every quarter, in one set week. I’ve since adapted it for mid-sized companies and run it with our customers. The short version:

1. Make it official. One week, hard deadlines, announced by the owner. Whatever misses the list waits for the next round, which is exactly why people take it seriously.
2. Talk to people one at a time. One or two people from each department with heavy operational work, never in a room together. In a room, the meeting turns into a debate about whose problem matters more. One-on-one, people tell you what is actually crushing them.
3. Problems first, solutions later. People describe what is wrong far better than they picture the fix. Solutions come in a second round, for the top problems only.
4. Put a number on every problem. The team estimates it, roughly: hours spent times what those hours cost, or how often something goes wrong times what it costs when it does.
5. Vote with a budget. At most 10% of a department’s list can be “fix now”. Asked on the spot, people name ten top problems. With a budget, they pick the handful.
6. Pick one thing and start it. The rest of the list is your roadmap.

The kit has the questions, the template and a prompt for each of these steps.

The conversations are where the expertise is, and it is already in your building.

“Won’t we have to redo it later?”

If you’re worried about that, the problem may not be acute enough to go first. You put a cost on it during the week, and a problem that costs that much every month pays for its own rework.

Building has also never been cheaper. What a bigger plan later replaces is the build. What carries over is the part that took the time: understanding the problem, and the data.

Run it yourself

We wrote the whole process up as a free kit: the week day by day, the questions to ask, a template for the problem list, and six prompts for any AI assistant to turn each conversation into rows and help you rank them.

If you bring someone in, bring them in to run the process, not to tell you what your business needs. That’s what our Speedrun is for.

Run the week, and at the end of it you’ll have a strategy, even if nobody calls it one.