Monthly AI board updates: what to include and leave out
By Keith Hamilton · · 4 min read
A monthly AI board update gives the board or owners a regular, one-page view of what AI is doing for the business. Too often that view is either missing or buried in a long slide deck full of activity. A good update is short, consistent and built around decisions. This guide sets out what to include, what to leave out, and how to keep it useful month after month.
What is an AI board update for?
An AI board update lets the board answer four questions quickly:
- Are we doing what we said we would do?
- Is it delivering what we expected?
- What is it costing, and are we within budget?
- What do we need to decide or watch?
These four questions are the core of board AI oversight. If a section does not help answer one of them, it does not belong.
How long should an AI board report be?
An AI board report fits on one page, with supporting detail available if anyone asks. A board that can read the update in five minutes will read it every month. One that runs to twenty slides will be skimmed.
Use the same layout every month. Consistency lets the board spot changes quickly and makes it obvious when something is missing.
What should a monthly AI board update include?
A monthly AI board update includes seven sections, in the same order every month.
1. Summary
Three or four sentences on the month: what moved forward, what results came in, and anything the board needs to know. Write it last, but put it first.
2. Progress against the roadmap
List each piece of work on your current plan with a simple status: on track, at risk or off track. For anything at risk or off track, give one line on why and what is being done. If you work to a quarterly plan, this is where the 90-day roadmap is tracked.
3. Results in pounds and hours
For each piece of work that is live, show the measure agreed at the start against its baseline. Report released time in hours and put a value on it using a stated hourly cost. Show where results sit against the low, central and high cases from the original AI business case.
Add one line on what each figure does and does not show. For example, a measure of calls answered shows whether calls were picked up, not whether the caller's query was resolved.
4. Spend against budget
Show what has been spent this month and to date, against budget, split into implementation, running costs and outside help. Include a note on internal staff time if it is significant.
5. Risks and incidents
Record anything that went wrong: incorrect outputs that reached a customer, data protection concerns, supplier outages or complaints. Note what was done about each. A short list of open risks, with an owner for each, completes the section.
Boards should expect to see some entries here. An update that never reports a problem is more worrying than one that reports small problems and how they were handled.
6. People
A line or two on staff: training delivered, how the team is finding the changes, and any concerns raised. AI changes how people work, and the board needs to hear about it.
7. Decisions needed
End with the specific decisions you need from the board this month, such as approving the next piece of work, extending a pilot or stopping something. If no decision is needed, say so.
What does a good results line look like?
A good results line states the baseline, this month's figure, the net saving, where it sits against the business case and where the released time has gone. Here is the format, filled in for an order-entry automation:
"Order entry from emails: baseline 10 hours a week. This month, 5 hours a week, including 1 hour checking AI outputs. Net saving 5 hours a week, worth about £109 a week at £21.82 an hour. This sits at the central case. Released time is being used to clear the returns backlog."
Replace the figures with your own. That level of detail is enough for the board to understand the result, check the arithmetic and see where the time has gone.
What should you leave out of an AI board update?
Leave out anything that does not link to an outcome or a decision:
- Activity counts with no link to outcomes, such as the number of prompts used
- Technical detail about models and settings, unless it affects a decision
- Supplier claims that have not been checked against your own figures
- New ideas that are not yet ready for a decision; keep these in a separate list
How does reporting AI to the board work across the year?
Reporting AI to the board works on a monthly and quarterly rhythm. The monthly update keeps the board informed. Once a quarter, hold a longer review: compare results with the plan, decide what to continue, adjust or stop, and agree the next quarter's priorities.
The monthly updates then become the evidence for that review, which gives you a simple, consistent form of AI governance reporting.
Where Forwardcycle fits
Reporting progress to the board in pounds and hours is a core part of our commercial leadership service. If you are at the start and want to know where AI fits your business, take the free AI Readiness Assessment.