How we work · 01
One decision.
Two to three weeks.
No code.
We work backwards from the decision to what would have to be true for it to go either way. Contracted so that “not this year” is a completed engagement rather than a failed one — which is what makes the finding trustworthy.
You hold: a written position, the evidence pack behind it and a ranked risk view.
What makes the finding worth having
A review that cannot come back negative was never a review.
The commercial problem with advisory work is structural: the firm writing the recommendation usually wants the implementation. Everyone involved knows it, so findings get read for what they are angling at rather than for what they say. We contract around it — the fee is fixed, the scope is fixed, and a written “do not do this” discharges the engagement completely.
We work from the decision backwards. Not “assess our AI capability” but “should we build this, buy it, or wait”. That naming happens before the engagement letter is signed, because a sprint pointed at a vague area produces a vague document.
And we talk to the people doing the work, not only the people sponsoring the change. That access is a condition of the engagement. The gap between how a process is described upstairs and how it runs downstairs is usually where the finding actually is.
Every finding carries an evidence state
Verified, Built, Designed or Intended — the same four words our own operation runs on. You can hold us to the document claim by claim, rather than being handed a confident summary with no way to test its parts.
How it runs
Name the decision, then work backwards from it.
Two to three weeks, fixed. Most of the value is in the first week’s access and the last week’s argument.
- Duration2–3 weeks, fixed
- FeeFixed, quoted in writing before work begins
- BuildsNothing — this shape produces a written position
- Access requiredThe people doing the work, not only the sponsors
- Honest exitA recommendation not to proceed completes the engagement
The decision, named in writing
What is being decided, by whom, and by when. If the answer is unclear, that conversation happens before the engagement letter rather than in week two.
Evidence gathered where the work happens
Interviews across functions and levels. Documents. Systems, where relevant. Self-declaration is treated as a claim to test, not as a finding.
The argument, both ways
What would make the decision go each way, and which of those conditions actually hold. This is the part that stops a document being a foregone conclusion in a suit.
Risks, ranked
Not a register — a ranking, with the reasoning. A risk list where everything is amber has told you nothing.
The written position
The recommendation, its alternative named, the evidence pack behind it, and an evidence state on every finding. Written to be read by a board and defended afterwards.
The two lists that matter
What you hold, and what we will not do.
Both are in the engagement letter before you sign it. The second list is the one worth reading twice — it is where most disappointment in this market actually comes from.
What you hold at the end
- A written position, with an evidence state on every finding
- The evidence pack behind it — not just the conclusions
- A ranked risk view, with the reasoning for the ranking
- A recommendation with its alternative explicitly named
- Whatever the specific service produces — portfolio, inventory, roadmap
What we won’t do
- Reviews that cannot name the decision they inform
- Maturity scores with no evidence behind the number
- Reviews where we cannot talk to the people doing the work
- Findings written to survive the sponsor reading them
- Advisory priced as the front end of a build we have already assumed
What is delivered under this shape
Five services run as an Advisory Sprint.
AI Opportunity Assessment, AI Readiness Assessment, AI Strategy & Roadmap, Build, Buy or Partner Review, and the AI Estate Review. Two of them are commonly run together in a single sprint.
See the Consulting Desk →Where the estate is the question
If the decision is about the AI you already run, a Sentinel Diagnostic takes a day and its fee is credited in full if you go further. Sometimes that is the cheaper first move.
Bring us the decision you keep circling.
Thirty minutes with a founder. No sales deck, NDA available beforehand. If a two-week sprint is the wrong shape for it, we will say so and tell you what the right one is.
Fixed fee · Quoted in writing before we start · NDA available