Build · Stage 3 → 4
Build
Ship assured AI applications you own.
Production-grade AI applications built to the Four Guarantees: measured, compliance-wrapped, agent-ready.
Your AI is proven and defensible, and the work now is putting more of it into production on that same rail.
What you get
- ✓ Choose your tier: Scaffolding (a repo you own, ships with the eval), PoC (trained and validated on your data, recommended), or Implementation (done-for-you, deployed, monitored)
- ✓ Each build ships an MCP endpoint + eval harness
- ✓ Drift monitoring + Annex IV documentation
- ✓ Document intelligence, forecasting, AML, CAIO, compliance and more
- ✓ Clinics and appointment-led businesses: predicting which bookings will not show, and turning that into a safe overbooking limit. Built with the patient's special-category data deliberately excluded and audited, not modelled.
The assurance ladder
Orient, Enable and Prove move you up the climb, but they carry no assurance grade. A grade only attaches to something built. Build is the first rung that ships an artifact, so it is the first that can be framed, then attested, then guaranteed.
A form. The harness, the eval scaffold and the gold-set slot, present and labelled, empty on purpose. The standard made portable.
Honest ceiling: "it's empty." The visible empty slots are the pitch for the fill.
A measurement. The harness filled against your data, a real calibration report: trust it here, abstain there.
Honest ceiling: "as of today." It will drift; the expiry is the pitch for the warranty.
A warranty. Monitored, drift-watched, standing assurance with defined recourse, the only state that can say "and it stays true."
Operational guarantee: defensible governance with recourse. Never legal sign-off or liability.
"Guaranteed" appears here and on Operate, nowhere below Build.
Scaffolding from €1,290 · PoC from €3,900 · Implementation from €12,000 + monitoring from €900 per month.
Every application is measured and compliance-wrapped, and shipped agent-ready for Stage 5.
Our methodology →