Skip to content

Approach

How an engagement actually runs

No discovery phase that produces only a slide deck, and no production phase that begins with throwing the prototype away.

  1. 01
    1 to 3 weeks

    Discovery

    Find the cases worth building and the ones worth killing.

    We interview the people who do the work, look at the data as it actually exists rather than as documented, and score candidate use cases on value, feasibility and risk. The output frequently includes a recommendation not to build something, which is cheaper to hear now.

    You end up with

    • Scored use case portfolio
    • Data and capability readiness assessment
    • Cost model per candidate
    • Recommended sequence with investment gates
  2. 02
    4 to 8 weeks

    Pilot

    Build the narrowest useful version and measure it honestly.

    One use case, real data, real users, instrumented from the first commit. We agree the success threshold before we start so the go or no-go decision is arithmetic rather than opinion. The pilot runs on the same architecture as production, because a prototype that has to be thrown away teaches you very little.

    You end up with

    • Working system in your environment
    • Evaluation harness and baseline scores
    • Measured cost and latency per request
    • Go or no-go recommendation against the agreed threshold
  3. 03
    8 to 16 weeks

    Production

    Harden, integrate and put it in front of everyone who needs it.

    Security review, access control, observability, failure handling, rate and cost limits, rollback. This is the phase most AI projects skip and most AI projects die in. We integrate with your existing identity, logging and deployment tooling rather than introducing a parallel stack.

    You end up with

    • Production deployment with CI/CD
    • Monitoring, alerting and cost dashboards
    • Runbooks and incident procedures
    • Governance evidence collected automatically
  4. 04
    Ongoing

    Enablement

    Hand it over properly, so you do not need us to keep it running.

    Documentation your team can act on, pairing sessions with the engineers who will own it, and a retained advisory line for the questions that come up in month three. We would rather be called back for the next thing than be a dependency on the last one.

    You end up with

    • Architecture and decision records
    • Engineer pairing and handover sessions
    • Retained advisory arrangement if wanted

The 6D Loop

What happens inside pilot and production

Stages two and three run on The 6D Loop. It is a loop, not a pipeline: Drive returns to Develop, because a system nobody revises is one nobody is measuring.

  1. 01

    Define

    Business question and success measures

  2. 02

    Discover

    Data sources, quality, ownership, access and contracts

  3. 03

    Develop

    Retrieval, models, orchestration and logic

  4. 04

    Decide

    Golden datasets, evaluations and quality gates

  5. 05

    Deploy

    Guardrails, release and rollout

  6. 06

    Drive

    Operations, cost, drift, quality and continuous improvement

Principles

Four commitments we will hold ourselves to

  • We will tell you not to build it

    A meaningful share of discovery engagements end with a recommendation against the project. That is the engagement working, not failing.

  • Measured, not asserted

    Every system we ship carries an evaluation harness. If we cannot measure whether it improved, we do not claim that it did.

  • Your engineers own it

    Code lands in your repositories, in your stack, with your team pairing on it. Lock-in is a business model, not an architecture.

  • Cost is a design constraint

    Inference spend is designed in from the first week, not discovered in the first invoice after launch.

Let's build together

Start with discovery

One to three weeks, a scored portfolio, and a straight answer about whether any of it is worth building.

Book a discovery callBrowse services