Four to eight weeks. Scoped deliverable signed off up front. Weekly demos, no surprises. CI/CD, observability, and evaluation harness included from day one. 30-day production warranty on whatever ships. From £45,000, fixed.
You have a clear AI feature to ship, a budget to ship it, and a team that is either not large enough or not deep enough in AI production to carry it alone. Typical shapes: a RAG system that has to reach faithfulness ≥ 0.85 to get past legal signoff, a multi-agent workflow that has to handle 10x production traffic by a specific date, a migration from a brittle prompt chain to a reliable LangGraph architecture with real evals.
Week 0 (pre-engagement): discovery call, written proposal in 48 hours. Week 1: architecture spike, data model, eval set definition, signed SOW. Weeks 2–6: building, with weekly demos and pull-request review cadence. Weeks 7–8 (if needed): hardening, load test, handover.
Deliveries scope themselves to the 4–8 week window. If the real ambition requires 12 weeks, we scope it as two sequential Sprints with a go/no-go gate between them.
Because T&M incentivises slowness. Fixed scope and fixed price means my incentive matches yours: ship well, fast. Genuine scope changes are handled in writing with a written amendment, before the next invoice.
Then the right first step is the Production Audit or a paid Discovery Sprint (one week, £8,500) where the deliverable is the SOW for the full Build Sprint. We tighten the scope until fixed pricing is safe for both sides.
Yes — that is the point. Your team owns the code afterwards. They pair on PRs, sit in on demos, and learn the patterns by doing. No "consultant leaves, nobody knows how it works" situations.
Book a 30-minute discovery call. If we are a fit, you have a written proposal in your inbox within 48 hours.
Book discovery call →