The item on the agenda
The chair has read about AI, a competitor has announced something, and at the last meeting someone asked: what is our AI strategy? It landed on your desk. The next meeting is not far away, and what you have is a scattering of staff using ChatGPT, a vendor or two who have pitched, and a sense that the board wants something ambitious.
The easy answer is a slide of tools and buzzwords. You know that will not survive the first sharp question: what will this actually change, what will it cost to find out, and what could go wrong?
Why AI plans come out vague
Most AI strategies are written from the technology outward. They list what AI can do in general and then look for places to put it. The result sounds impressive and is hard to act on, because it does not connect to how your business actually makes money or where it loses time.
The other common gap is data. Many good AI ideas depend on information that is scattered, incomplete or locked in systems that are hard to get at. A plan that ignores this promises things the business cannot yet deliver, and the board finds out a year later.
What a weak plan costs
| Weak plan | What follows |
|---|---|
| A list of tools | Licences bought, little change in how work is done |
| Too many initiatives | Effort spread thin, nothing reaches production |
| No data assessment | Projects stall when the data turns out not to exist |
| No risk section | Data protection or client contract issues surface late |
| No measures | Nobody can say whether it worked, so funding stops |
There is also a cost to doing nothing credible. Staff are already using AI informally, and without a plan that behaviour carries on unmanaged.
How we help you build the plan
- We interview heads of department and a few front-line staff to find where time goes on repetitive work, where decisions wait on information, and where errors cost money.
- We look at what AI is already being used for, officially or not, since that shows where demand is real.
- We assess the data behind each opportunity: where it lives, how complete it is, and whether it can be reached through APIs from systems such as Microsoft 365, Xero, Salesforce or your ERP.
- We score opportunities on value, feasibility and risk, and we are explicit about the ones that are not ready and why.
- We set out the guardrails: an AI usage policy, how client and personal data will be handled, who approves new AI tools, and how outputs are checked.
- We shortlist a small number of first projects, each with a clear owner, a measure of success agreed in advance, and a decision point for scaling up or stopping.
The output is a short board paper and a supporting appendix, written for directors rather than technologists. We can present it with you, and answer the technical questions so you do not have to field them alone.
What the board gets
A plan that starts from the business: here is where we lose time and money, here is what AI can realistically do about it with the data we have, here is what we will try first, what it will need, and how we will know. Also: here is what we are deliberately not doing yet.
For you, it means the next meeting is a discussion about priorities rather than a defence of a slide of logos.
It also gives the board something to hold you to that is fair. Each first project has a measure agreed before it starts and a point where the business decides whether to scale it or stop. Stopping a project that did not work becomes a normal outcome of the plan, not an embarrassment that gets buried.
Is this your situation?
- The board or owners have asked for an AI strategy and you are holding the pen.
- You have vendor pitches but no independent view of what fits.
- Staff use AI informally and there is no policy or register.
- Nobody has assessed whether your data can support the ideas being discussed.
- Previous technology plans produced lots of activity and little change.