Skip to content
AI Board

For strategy managers

AI for strategy managers: know whether the plan is landing, before the MT asks

It is the evening before the MT update, the OKR file is open, and you are deciding what to lead with. Writing the strategy was never the hard part: knowing whether it is landing is. An AI grounded in your own progress files, QBR numbers and approved business cases answers that question with the source named, so you walk in with an escalation instead of a colour chart.

First, a disambiguation, because the search results confuse the two constantly. AI for strategy managers is not the same as an AI strategy. An AI strategy is a plan you write about AI: where the company should apply it, what to buy, what to govern. AI for strategy managers is AI pointed at your own job: the annual plan, the OKR progress pack, the QBR deck, the follow-through on initiatives the board already approved. This page is about the second one.

The distinction matters because the first half of the job is already served. BCG's Henderson Institute found that AI tools and agents have delivered consistent, positive impact for strategy leaders only in market intelligence and research, while the judgment-intensive work (M&A, partnerships, portfolio choices) has not yet seen material improvement. Desk research is solved. Follow-through is not, and follow-through is where your week actually goes.

That gap has a measured size. PMI's study of more than 5,800 project professionals found just half of projects meet a modern definition of success, with 13% failing outright and 37% only partially delivering; in an accompanying survey of 300 senior executives, the most-cited barrier to reinvention (35%) was a disconnect between planning and delivery. Every tool on this market helps you write the plan. Almost none of them tell you whether it is landing, in the files where the evidence already sits. AI Board reads those files as they are (Strategy_OKR_Progress_Q3_2026_v4.xlsx, the QBR reconciliation, the approved business case PDF from two Octobers ago), with no migration into an OKR platform and no waiting on an integration. Every figure comes back with the document and version it came from, so you can re-check it in seconds before it goes in front of the MT. For the mechanics of asking, see chat with your data.

The week you recognize

Winging the MT update because you cannot tell what is actually on track

The update is tomorrow, nine objectives are in the file, and the honest answer to "are we on track" is that you would need two hours of cross-reading to know. So you present the colours, lead with the green, and hope nobody probes the two amber ones. The thing that costs you is not the missing number: it is that the two slipping objectives usually slip on the same quiet assumption (an unfilled commercial hire, a delayed pricing decision) and nobody restated it, so the MT gets a status report where it needed an escalation with an owner and a date.

Nobody remembers what the approved initiative actually promised

A year after approval, someone asks whether the initiative delivered. The business case that answers it is a final PDF in a SharePoint folder nobody opens, and the assumptions behind its numbers have quietly left the building with the people who wrote them. So initiatives get written off as failures when they were behind on adoption, not on the case, or defended as successes because nobody can prove otherwise. The distinction between a rollout problem and a case problem is the one thing that would change the decision, and it is the one thing that is no longer at hand.

An objective that only looks missed, and the number nobody will reconcile in time

Sales and finance report different revenue, so the growth objective in your QBR deck looks missed when the gap is pure timing across a handful of named deals. You inherit it the moment it hits the report, and you are the one who has to decide which figure is defensible. What you need is not the accounting answer (finance is almost always right here) but the bridge back to the objective: the target is on track, the recognition is in the next quarter, and the footnote says so in one line.

Live demo

Your personal AI assistant, thinking

What did project P-2043 actually deliver against the business case we approved?

A year in, tracked benefit is 60% of the EUR 300k savings the approved case promised, behind on adoption, not the case. Reopen the original business case and fix the rollout, don't write it off.

Business_case_P-2043_approved.pdfapproved Oct 2024 · final
Ask AI Board…

What changes

Lead with what is off, with an owner attached

Ask what is actually on pace and you get the objectives that are slipping first, the shared assumption underneath them, since when it has been true, and who owns it, read straight from the progress file with its version named. You still decide what to do about it. What changes is that you open the MT with the two objectives that need a decision rather than the six that need applause, and you can point at the file when someone pushes back.

The approved business case, live again

Ask what an initiative delivered against the case that was approved and you get the promised benefit, the tracked benefit, the gap, and whether the gap is adoption or the case itself, with one concrete lever named (which department, which usage is lagging) so it becomes something you can assign on Monday. The approved PDF stops being an archive and becomes a document you can interrogate a year later, which is the whole point of having approved it.

A reconciliation that lands back on the objective

Ask which revenue figure belongs in the deck and you get the defensible number, the named deals causing the timing difference, and the sentence that connects it back to the objective it appears to miss. Not an accounting answer handed back to you: the strategic reading of it, in a form you can footnote. Ten minutes of chasing the controller becomes one line in the deck that survives the first question.

Answered on demand

Board deck evidence

Which figures in this board deck cannot be traced back to a source file, and which version did each of the others come from?

Assumptions that expired

Which assumptions in last year's strategy paper no longer hold given what the current numbers show?

Initiative follow-through

Which approved initiatives have had no progress update in over a quarter, and who is named as owner on each?

Trade-off before the planning round

If we have to drop one objective this quarter, which one is least load-bearing for the margin target?

Questions, answered

Is AI for strategy managers the same as an AI strategy?
No, and the difference is the whole page. An AI strategy is a document you produce about AI: where the company applies it, what to buy, how to govern it. AI for strategy managers is AI applied to the strategy job itself: tracking progress against the annual plan, reconciling numbers between functions before the QBR, checking what an approved initiative actually delivered, and finding the assumption behind an objective that is slipping. If you searched for the first thing, most results will serve you. If you searched for the second, almost nothing on the market is written for you, which is why this page exists.
Can AI track progress against our OKRs or annual plan without moving into another platform?
Yes, and that is the design choice that matters. The usual answer is to migrate your plan into an OKR or strategy-execution suite and wait for the integrations; your update is due Thursday and the evidence is already in a spreadsheet you maintain by hand. AI Board reads the file as it is, names the document and version behind each figure, and leads with what is off rather than what is green. It is private by design: it runs grounded in the files on your own machine, which is why board material and pre-announcement numbers can go near it at all. See [security](/en/security) for how that holds up.
Can AI tell me whether a project delivered its approved business case?
It can compare what the approved case promised against what is being tracked now, show you the gap, and separate a rollout problem from a case problem, all from the approved PDF and the current numbers. That question matters because so much of what gets started never gets checked: [MIT's NANDA research reported that around 95% of enterprise generative-AI pilots delivered no measurable P&L impact](https://fortune.com/2025/08/18/mit-report-95-percent-generative-ai-pilots-at-companies-failing-cfo/). That figure is contested, on a small sample and a loose definition of failure, and we would rather say so than quote it flat: checking the assumption behind a number before you repeat it is exactly the habit this page is about.
Will AI replace strategy managers?
No, and the honest evidence points the other way. [HBR's case studies of CEOs using generative AI for strategic planning](https://hbr.org/2024/09/how-ceos-are-using-gen-ai-for-strategic-planning) found it surfaced challenges and questions management had missed and helped counter human bias, but it was weaker at imagining genuinely new scenarios, because its reasoning is rooted in what has already happened, and it missed possibilities specific to that company's own capabilities. Which is the shape of the tool: it reads more than you can, remembers what the plan assumed, and never gets bored of the follow-up. The judgement call in front of the MT stays yours.

The grounding is the part people skip. Only 7% of enterprises say their data is completely ready for AI, which is why a generic chatbot gives you fluent strategy prose and no idea whether your plan is landing. AI Board is your personal AI assistant that makes you AI-native: it runs on your own laptop, grounded in your company's data, and gets sharper as your knowledge grows. It answers from your files, names the source every time, and never starts at the green.

Put your own strategy files to work

See what happens when your OKR progress file, QBR numbers and approved business cases can be asked a direct question, private by design, with the document and version named every time.

Runs on your own laptop. Your data never leaves it.