Skip to content

    AI Pilot Build

    6–12 weeks to turn a scoped use case into a working pilot — with named exit criteria and no lock-in.

    If you've finished a Scoping Lab (with us or someone else) and the answer is "build it," the Pilot Build is the next step. We build the system, integrate it where it needs to live, and hand over the code, the documentation, and the runbook. You own everything.

    Discuss a Pilot Build

    Who it's for

    • Teams with a scoped use case and a clear success metric — not an idea looking for shape
    • Organisations that want a working pilot in production, not a Jupyter notebook on a consultant's laptop
    • Buyers who want a fixed-price, fixed-scope build with a clear exit — no retainer, no lock-in

    What's included

    • Working pilot built to the success criteria agreed in scoping
    • Integration into your environment (cloud, data sources, identity, monitoring) — not a sandbox demo
    • Evaluation harness: how you'll measure the pilot in production, not just in development
    • Human-in-the-loop design where the use case requires it — not bolted on at the end
    • Source code, infrastructure-as-code, documentation, and runbook — handed over, not retained
    • Knowledge transfer to your engineering team or chosen build partner
    • AI Act risk classification, GDPR review, and audit-ready logging
    • Two weeks of post-handover support, then a clean exit (or a separate optimisation engagement if you want one)

    Duration: 6 weeks (well-scoped use case, clean data, single integration) to 12 weeks (multi-system integration or stricter compliance needs).

    Investment: Typically €40–80k, fixed price per pilot, scoped from the Scoping Lab output. We quote before we start.

    Deliverable: A working pilot in production, fully owned by you, with everything needed to run it without us.

    What this is not: a long-running implementation programme. The Pilot Build ends when the pilot is in production and your team can run it. If you want us to keep building, that's a separate conversation with separate scope.

    What you walk away with

    Not a document — a working system you own. Handed over:

    • The working pilot, running in your environment
    • Source code and infrastructure-as-code, in your repositories
    • Evaluation harness, documentation, and a runbook
    • AI Act risk classification and audit-ready logging