Independent AI engineering consultancy

AI engineering for teams with established codebases.

We build AI into the systems you already run. Fixed price, fixed duration, one agreed outcome metric. You own everything we build.

30 minutes · free · no pitch

How the harness fitsAny model feeds into a harness made of an evaluation suite, a context pipeline and review gates, which ships into your codebase through CI. Results measured in production feed back into the evaluations.ANY MODELTHE HARNESSYOUR CODEBASEClaudeGPTGeminiOpen weightsscored on your evalsEvaluation suitemeasured, not eyeballedContext pipelinerepo map · retrieval · toolsReview gateshuman where it mattersCI integration├─ src/│ ├─ billing/│ ├─ legacy/│ └─ api/├─ ci/│ └─ evals.yml├─ docs/└─ …10 yrsproductionyour instrumentationmeasured in prod → back into the evals
  • In production at a network orchestration platform
  • Toolchain migration for a thirty-year-old codebase
  • Retention, role-based access, published subprocessors

01 / Offers

What we sell.

Start with the free half hour. Everything after it has a price and a duration.

  • You want an honest read before spending anything

    30 minutes on your actual problem. No pitch, no deliverable.

    30 min

  • Teardownfrom $4,500

    Leadership wants a number for AI and you don't have one

    A prototype agent on your code, a written assessment, a 90-day plan

    2–3 weeks

  • Harness Sprintfrom $28,000

    Your team is slow on an old codebase and the AI tools didn't stick

    Evaluation suite, context pipeline, review gates, CI integration

    6–8 weeks

  • Build Sprintfrom $30,000

    You need one AI capability shipped into a system you already run

    The capability in production, with evaluations and a human review step

    6–10 weeks

  • Embedded$9,000–24,000/mo

    You want the engineers who built it to stay on

    Forward-deployed engineering, one to three days a week

    3-month min.

Fixed price, never hourlyYou own everything we buildModel-agnosticHalf the Teardown credits toward a sprint

02 / What we believe

How well a model performs depends on the system around it.

The same model produces very different results depending on that system. We build that system.

  • Evaluation suite
  • Context pipeline
  • Review gates
  • CI integration

03 / Why us

What sets us apart.

  • Production experience in mature codebases
  • Independent of model vendors
  • Fixed price, fixed duration, measured outcome
  • The people who sell the work do the work

Start a conversation

Book office hours.

30 minutes, no pitch. We look at your actual problem.