Four phases, and you can stop after any of them
Most AI engagements fail because nobody agreed what success would look like. We make that the first deliverable.
What each phase produces
No phase depends on you committing to the next one.
Discovery
1–2 weeksWe map the problem, the data, and the constraints. You get a written technical assessment with a recommended architecture and a cost model — useful even if you stop there.
- Technical assessment
- Architecture proposal
- Cost + latency model
- Risk register
Prototype
2–4 weeksA working slice of the real system against your real data. Not a mockup. We measure retrieval quality and answer accuracy before committing to a full build.
- Working prototype
- Evaluation harness
- Quality baseline
- Go / no-go recommendation
Build
6–16 weeksTwo-week increments, demoed live. Your team has repository access from day one. Every increment is deployable, tested and documented.
- Production system
- Test suite
- CI/CD pipeline
- Runbooks
Operate & hand over
OngoingMonitoring, evaluation dashboards and on-call during stabilisation. We train your engineers to own it, then step back to whatever support level you want.
- Observability stack
- Eval dashboards
- Team enablement
- Support agreement
How we actually behave
Evaluation before opinion
We do not claim a change improved quality until a measurement says so. Every engagement builds an evaluation harness early, because it is what makes the rest of the work honest.
Your repository, from day one
Your engineers have access to everything we write, as we write it. No big reveal at the end, no code we are protective of.
Two-week increments, demoed live
Every increment is deployable and shown working against real data. Progress you can see beats progress you are told about.
We will tell you not to build it
If the assessment says the value is not there, that is what the document says. It has cost us work and earned us clients.
Handover is the goal
We succeed when your team can extend the system without us. Documentation, runbooks and pairing are part of delivery, not an upsell.
Fixed scope, honest change
Phases are fixed-price where scope allows. When something changes we say so immediately rather than absorbing it quietly and resenting it later.
Ways to work with us
Assessment
1–2 weeksA written technical assessment with architecture, costs and risks. Standalone and useful on its own.
Best when: You are deciding whether to invest at all.
Prototype
2–4 weeksA working slice against your real data with measured quality, ending in a go / no-go recommendation.
Best when: You need evidence before committing budget.
Full build
6–16 weeksAn embedded team taking a system from design to production, with handover built in.
Best when: You know what you need and want it shipped.
Embedded team
OngoingEngineers working inside your team on your roadmap, with our research bench behind them.
Best when: You need sustained capacity, not a project.
What clients say afterwards
“They rebuilt our retrieval layer and our support deflection rate went from 19% to 54% in six weeks. The difference was grounding, not a bigger model.”
“The only vendor that showed us their evaluation numbers before asking for a build budget. That bought a lot of trust.”
“We kept the team on after launch. They write the kind of code our own engineers wanted to inherit.”
Let's talk about what you're building
Tell us the problem. We'll tell you honestly whether AI is the right tool, and what it would take.