How we work

Three moves. One team. No handoffs.

Scope, Build, Operate — the same partners and engineers from first conversation to the year of operation that follows go-live. No "delivery team" you've never met. No retainer-driven roadmaps. One team, accountable end-to-end.

01 of 03
Scope.
≈ 2 weeks

A short, sharp engagement to pressure-test the thesis, the data, and the appetite. We treat scoping as a standalone deliverable — not a free pitch.

What happens in two weeks

  • Stakeholder interviews across business, technology, and operations.
  • A data assessment — does the foundation actually support what's being proposed?
  • Pressure-testing the use case against six common failure modes we've learned to look for.
  • A written Scope memo with our honest call: build it, change the scope, or don't build it.

What you get

  • A defensible Scope memo — board-ready, written, yours to keep.
  • An architecture sketch sufficient for budgeting and team-sizing.
  • A cost-and-capability envelope: what it would cost, what it would do, what it wouldn't.
  • A clear recommendation on go / no-go / re-scope.
The honest part

We turn down work we can't ship. Roughly one in three Scope engagements ends with a "don't build this" recommendation — and we still hand over the memo. That's the deal.

↘ no free pitches.
no sandbagging.
02 of 03
Build.
8 – 24 weeks

Production-grade engineering. The same team that scoped it ships it — partners and engineers stay on the engagement together, so the people who promised the architecture are the people writing the code.

What we build

  • The model, the pipeline, and the platform around them — production-deployed, not notebook-grade.
  • An evaluation harness that runs continuously, not just at the milestone demo.
  • Observability and cost telemetry from day one. We instrument what we plan to operate.
  • Governance hooks for the parts that need to be audit-defensible: data lineage, access control, model cards.

How we work during build

  • Weekly working-session cadence with the client team — not status decks.
  • Open-by-default documentation: anything we'd hand over at the end gets written as we go.
  • Code in the client's repos, not a black-box delivery handed over at the end.
  • A clearly-defined "shippable" definition agreed at the start — and held to.
The non-negotiable part

"Shippable" means it can go to production. Not "the demo worked." Not "it passed UAT in a sandbox." Production — observed, evaluated, and reversible if it breaks.

↘ same names from
scope through ship.
03 of 03
Operate.
6 – 24 months

We stay. The part that decides whether AI works a year from now — drift, retraining, cost, latency, evals — is the part most consultancies don't sign up for. We do.

What "Operate" actually covers

  • Continuous evaluation against the agreed shipping definition. If it stops being shippable, we know first.
  • Drift detection on both data and model — with retraining when it's warranted, not on a fixed cadence.
  • Cost watch and latency watch — we own the operating envelope we agreed to.
  • Incident response for AI-specific failure modes that traditional ops teams aren't trained for.
  • Quarterly written reviews — what's running, what's changed, what's at risk.

How the engagement ends

  • We hand back to your team when they're ready, not when our retainer ends.
  • A written runbook covering everything we've learned about operating this specific system.
  • A transition period co-running operations with your team, so the muscle memory transfers.
  • An open invitation back if something material changes — but no automatic renewal clause designed to trap you.
The whole point

Most AI engagements end at "deployed." Ours end at "your team can run this without us." Until that's true, we're still on it.

↘ this is the rare verb.
Commercials & engagement structures

How we charge — published.

We publish our engagement structures and rates rather than hiding them behind "let's talk." If the maths doesn't work for your budget on this page, the maths won't work behind a sales call either.

Scope sprint

Fixed-price, two-week engagement. One partner plus one engineer. Deliverable is the Scope memo and architecture sketch. Priced to be standalone-viable — you can take the memo and use a different team for the build if our call says so.

Build engagement

Time-and-materials with a clearly-defined "shippable" milestone and a cost envelope agreed at Scope. Partner stays on. Team scales with the work — usually two to five engineers. Weekly burn reports.

Operate retainer

Monthly retainer covering the operational obligations above, sized to the system. Three-month minimum, no auto-renew. We bias toward your team owning the system within twelve months.

Workshops & advisory

Fixed-fee executive workshops, capability assessments, and board-level advisory days. Useful when the question is "do we even need this?" — not just "build it." We deliver and leave; no follow-on sales pressure.

// Detailed rate card available on request · No NDA required for the rates themselves

The promise this all rests on

Same team. Scope through operation. Always.

— BridgeOne operating principle 01