I have read a lot of AI strategies. Most are carefully made, extensive and practically worthless, because they fail to answer one question: what do we do first, and what do we deliberately not do?
A strategy is a choice. A document that lays out thirty use cases as equals has not made that choice. It has delegated it to an organisation that cannot make it.
What belongs on the two pages
Five points are enough, and each can be stated in a few sentences.
- The position: what do we use AI for, and what do we explicitly not use it for? One sentence each.
- The three initiatives for the next twelve months, each with one sentence on the expected result, in measures the business already tracks.
- The three things we need to be able to do for that and cannot do today.
- The guardrails: what is allowed, what needs approval, what is ruled out.
- The checkpoints: how will we know at six and twelve months whether it is working, and who is looking?
Everything else is appendix. The appendix may run to sixty slides, but it is not discussed in the executive meeting.
You recognise a strategy not by what is in it, but by what was deliberately left out.
Saying no is the most valuable part
The hardest and most important section is the list of what will not be done. It is unpopular, because it disappoints someone, usually a committed person who proposed a use case.
But without that list, every initiative competes for the same scarce resources: the attention of the leadership team, the time of the business units, and the few people who understand both the business and the technology. Those resources are the real bottleneck, not the budget.
How the two pages come about
Not through a strategy document, but through conversations. You need a solid picture of how value is created, an honest assessment of your own capabilities, and a set of priorities the leadership team has agreed together. Four to eight weeks is realistic.
The result is shorter than what most consultancies deliver. That is not a lack of substance. Brevity is the product of decisions. Length is often the product of avoiding them.
The test
Give your strategy to a manager who was not involved in writing it, and after five minutes of reading ask: what are we doing first, and what are we not doing? If the answer comes hesitantly, the problem is not the person.