Example scenario, not a client project. A SaaS company has a dozen or so "AI" ideas in its backlog from different departments. At a two-day workshop, the team scores them by value to the customer, data quality and risk of error. Three remain. One — the most impressive — is consciously set aside, because it would need data the company doesn't have.
Is this for you?
- You have a list of 20 AI ideas and a budget for two.
- A pilot has stalled: it works technically, but no one knows whether or how to go further.
- The board wants an "AI strategy", and the team needs a concrete next step.
- The product vision is general, and everyone understands it differently.
What I do
- Product strategy workshop — one or two days with decision-makers, ending in priorities.
- A short design sprint — from the problem to solution sketches, and choosing one to test.
- Clarifying the vision — from a general "we want AI in the product" to concrete use cases.
- Assessing a stalled pilot — what the pilot really showed, what's blocking it, and whether to save it, change it or close it.
The AI layer
- For every idea we settle which decisions stay with a person and which can be automated.
- We score ideas by value, feasibility with the data you have, and risk of error — not by the "wow" effect.
- With pilots, we ask about what usually kills them: no learning loop, no trust, no process owner (MIT NANDA 2025; Gartner: over 40% of agentic AI projects will be canceled by 2027).
- Context: according to BCG 2026, the advantage comes from strategy and from changing how work runs, much less from the tools themselves.
What you get
- An opportunity map — ideas scored by value, feasibility and risk.
- 3 priorities, with the reasoning and an owner.
- A decision log — what we do, what we don't, and why.
- A first-step plan — what we test, how, and how we'll recognize the result.
- For a pilot assessment: a recommendation — continue, change or close.
How it runs
- Preparation — conversations with participants, gathering ideas and data.
- Workshop or sprint — 1–5 days, depending on the format.
- Write-up — the map, the priorities and the decision log.
- Closing session — confirmation with the decision-makers.
Roughly 1 to 3 weeks [to be confirmed].
How we'll know it worked
- Decision-makers sign off on the same 3 priorities.
- The team knows what it's doing in the coming weeks.
- At least one idea was consciously rejected — and everyone knows why.
What this doesn't cover
I don't create company strategy or business plans. I work at the level of the product and the process: what we change and how we'll test it.
Visualizations
Opportunity map (sample artifact)
Sample artifact: ideas scored, three chosen.
Assessing a stalled pilot
A stalled pilot needs a decision, not another month.