Home · Method
Our Method

Proof-First Build.

Most AI and software projects don’t fail on technology. They fail on unclear objectives and unproven assumptions — discovered after the budget is spent. Proof-First Build reverses the order: evidence first, commitment after. You see the system working on your own data before you approve the full build.

Step 01Fit call

30 minutes

A conversation, not a pitch. You describe the problem as it actually is; we ask the questions that decide whether a build makes sense at all. About a third of the time, the honest answer is that you don’t need us — a configuration change, a different tool, or a hire first. We’ll tell you which.

  • A straight answer on fit, on the call
  • If it fits: what a Blueprint would examine
  • If it doesn’t: who or what to try instead
Step 02Blueprint

Free · about a week

We examine the problem properly before anyone commits to a build. Your data, your systems, your constraints — reviewed hands-on, not from a questionnaire. The output is a written plan you keep, whatever you decide next.

  • Architecture for the proposed system
  • Data review: what’s usable, what’s missing, what’s risky
  • Success metrics, agreed in writing
  • Risks and unknowns, named explicitly
  • A fixed-scope proposal for the pilot
Step 03Pilot on your data

30–60 days

A working system on a bounded slice of your real operation — one team, one document set, one workflow. Not a demo with sample data: the real thing, at a scale where mistakes are cheap and learning is fast. Weekly demos, and measurement against the Blueprint’s metrics from week one.

  • Working system on real data and real users
  • Evaluation results against agreed metrics
  • A go / adjust / stop recommendation with evidence
Step 04Production & handover

You own it

The pilot, hardened: security review, monitoring, documentation, and rollout to the full operation. Everything lives in your cloud accounts and your repositories. We stay through a support window after launch — then the system is simply yours.

  • Production deployment in your accounts
  • Documentation and a team walkthrough
  • Monitoring and evaluation running from day one
  • Support window after launch
The Free Blueprint

What’s inside it.

  • Problem statement — the outcome you’re buying, in one paragraph
  • Architecture — the system design, and why it’s shaped that way
  • Data review — what your data can support today, and what it can’t
  • Success metrics — the numbers the pilot will be judged against
  • Risks — the ways this could fail, named before they’re expensive
  • Fixed-scope proposal — the pilot’s scope, timeline, and terms in writing
FAQ

Questions about the method.

Why is the Blueprint free?
Because it protects both sides. You get a real plan you can act on with anyone. We get engagements scoped on evidence instead of optimism — which is why our pilots tend to reach production. If the Blueprint says don’t build, believing it is the cheapest outcome available.
What if the pilot misses its metrics?
Then the evidence says adjust or stop, and we say so — with the evaluation results in front of both of us. A stopped pilot costs weeks. A failed full build costs quarters.
Do you work any other way?
For advisory engagements, yes — a few hours a week, no build attached. For builds, Proof-First Build is the way we work; it’s the reason clients end up owning systems that survive production.
Who owns the Blueprint and the pilot?
You do. The Blueprint is yours to keep from the moment it’s delivered, and everything built in a pilot lives in your accounts under your ownership.

Start with a fit call.

Thirty minutes. Bring the problem as it actually is — the stalled pilot, the bill, the spreadsheet everyone quietly maintains. We’ll tell you honestly whether we’re the right builder for it.

You leave with a straight answer on fit — and if it fits, a free Blueprint to follow.