AI strategy & agents
Know where AI pays for itself, and where it will not.
Most AI advice comes from people who never have to run the result. We build and operate AI inside production platforms, which mostly teaches you where not to put it — so you get a decision you can defend to a board and to an engineer in the same meeting.
What you get
What you are left holding.
The value should not leave when we do.
- A written assessment with the reasoning, not only the conclusion
- An opportunity map scored on value and feasibility, in your terms
- A data-readiness verdict per opportunity, including the ones that fail it
- A written decision record for anything we recommend building
- Working code and its tests, where the engagement includes building
- A named list of the things we advise you not to do, and why
How it goes
Agents do the work. People decide.
This is not a slogan; it is the way we run our own product, and it is the part worth copying.
Everything is written first
A decision that is not written down was not made. What to build and why comes before the code and outlives whoever wrote it.
01
Agents do the execution
A large share of the work is done by agents working from those documents, inside a harness that limits what they are allowed to touch.
02
The checks are machine-run
Tests and guards run before anything can merge, and a check that could not have failed is treated as broken rather than as good news.
03
A person approves
Nothing reaches a running system without a human review that could have said no. Your engagement gets the same gate.
04
Which one is you
Three shapes, one team.
AI strategy and audit
For a leadership team that has to decide this quarter. You get an opportunity map scored by value and by feasibility, a verdict on whether your data can support each one, a build-or-buy answer, and a roadmap that accounts for the people who have to use it.
2–3 weeks
Agents and automations
For a team that already knows what it wants automated. Built on your systems or on ours, with typed boundaries, evaluation before rollout, an approval step wherever an agent acts on someone's behalf, and enough visibility to answer what it did and why.
scoped per system
Fractional CTO
For a founder who needs the decisions held as well as made. The delivery model we run internally, applied to your product: architecture written down, agents doing the execution, people holding the gates, every change reviewable by somebody who was not in the room.
monthly, ongoing
Questions
What you will want to ask.
- How long does an audit take?
- Two to three weeks from the first call, and the written assessment is yours whether or not you build anything with us.
- Do you build on our stack or on yours?
- Yours when what you have is worth building on, ours when starting there would cost you less than repairing what you have — and we tell you which before you commit.
- What if the honest answer is that we should not use AI here?
- Then that is the deliverable, named and argued, and being willing to say it is most of what you are paying for.
- Can we start with one and continue with another?
- Most people do: an audit that finds something worth building usually turns into building it, and a fractional arrangement usually starts after both.
AI strategy & agents · next step
Send the question.
Even the awkward one. Especially the awkward one.