A plan specific enough to act on, including when the answer is don't build it.
AI strategy, architecture reviews, stack evaluation and build roadmaps. You get a written recommendation with a sequence, a cost and the risks named — useful whether or not we are the ones who build it, and priced separately so that stays true.
For leadership teams, CTOs and founders across the USA, UK and Europe. One to three weeks.
Four questions worth answering before you spend.
AI opportunity audits
Which workflows would genuinely pay for automation, and which only demo well. Usually the answer is the boring high-volume process nobody wanted to look at.
Architecture reviews
An honest read of a system you inherited, outsourced or outgrew: what is sound, what is fragile, what is worth keeping, and what it costs to fix versus replace.
Roadmaps and sequencing
What to build first and why, with dependencies made explicit. Order matters more than scope — most failed programmes built the right things in the wrong sequence.
Vendor and model selection
Build, buy or assemble, and which provider. Independent, because we do not resell anyone's platform and have nothing to gain from the answer.
Look properly, write it down, talk it through.
Look properly
Interviews with the people doing the work, a read of the systems, and the numbers behind the process. One to three weeks, depending on how many systems and stakeholders are in scope.
Write it down
A document specific enough to hand to an engineering team or a board: recommendation, sequence, cost, risks, and what to do first.
Walk you through it
A session with the people who would do the work, so the reasoning survives contact with questions rather than sitting unread in a shared drive.
Built for the decision before the build.
- →Leadership teams told they should be doing something with AI and unsure where to start
- →CTOs who inherited a system nobody can confidently describe
- →Founders choosing between building, buying and assembling
- →Teams whose last technology programme quietly failed and want to know why
We have nothing to sell you here.
This engagement is priced on its own rather than folded into a build, because an assessment that only pays off if it recommends a project is not an assessment. A fair number end with an off-the-shelf tool, a smaller scope than expected, or a recommendation to wait.
A document, not a deck.
a document you can act on without us
Recommendation, sequence, cost, riskswe resell nobody's platform
Including when the answer is do not buildRelated field notes and proof.
Custom software development
When the roadmap says build, the people who wrote it are the people who ship it.
From pilot to production
The four reasons AI projects never leave proof-of-concept, and the checklist that gets them out.
Build versus buy, honestly
The comparison we run before recommending a build, including when a subscription wins.