From first conversation to production in eight weeks.

Every engagement follows the same four stages, each with a clear exit. The Diagnose brief is yours to act on whether or not we build what it recommends.

  1. 1

    Diagnose

    Weeks 1–2

    An audit of the workflow, the data and the systems it touches, ending in a written brief you can act on with or without us.

    You receive

    • Workflow and data audit
    • Opportunity map of the top three agents
    • Success metric definition
    • Build estimate and risk log

    Security. Data map and threat model for the first agent.

  2. 2

    Prototype

    Weeks 3–4

    A working agent on your data in a sandbox, tested end to end by your own team before anything is hardened.

    You receive

    • Sandboxed agent on your data
    • Evaluation harness on your ground truth
    • Baseline against the agreed metric

    Security. Runs isolated from production, on data you approve.

  3. 3

    Integrate

    Weeks 5–7

    The agent wired into your production stack, with permissions, guardrails, approval gates and observability.

    You receive

    • Production integration
    • Guardrails and approval gates
    • Monitoring and rollback
    • Full source, prompts and runbooks

    Security. Red-team pass before release; scoped permissions go live.

  4. 4

    Operate

    Week 8 onward

    Weekly evaluations against your ground truth, retrospectives and a shared roadmap. We stay until the metric moves.

    You receive

    • Weekly evals and retros
    • New capabilities on a rolling roadmap
    • 24/5 SLA response on managed operations

    Security. Re-tested on every release; audit trail reviewed with your team.

Three ways to work with us.

Most clients start with a Diagnostic and move into a Build when the numbers make sense. Commercial terms are scoped per engagement and shared in your briefing.

  • Diagnostic

    2 weeks

    The right thing to build, in writing. A brief you can act on with or without us.

    • Workflow and data audit
    • Opportunity map of the top three agents
    • Success metric definition
    • Build estimate and risk log

  • Build

    8 weeks

    One production-grade agent, wired into your stack and owned by you.

    • Everything in the Diagnostic
    • Custom agent and evaluation harness
    • Full source and prompts handed over
    • Deployed to your infrastructure
    • 60 days of hypercare

  • Operate

    Ongoing

    We run and evolve your agents alongside your team.

    • Managed operations
    • Weekly evaluations and retros
    • New capabilities every month
    • 24/5 SLA response

Questions buyers ask first.

Whose data touches the models?

By default, models run inside your VPC. Where a hosted model is the right choice, it runs on a zero-retention tier under your own key. Your data is never used for training.

What if the agent does not hit the metric?

We agree the target metric in Diagnose and do not ship until an evaluation on your ground truth clears it. If we miss, we keep iterating at our own cost.

How are engagements priced?

Every engagement is scoped to a fixed fee, agreed before work starts and shared in your briefing. No hourly billing, no equity, no revenue share.

Can our own team take over later?

Yes. We help teams stand up an internal AI function and roll off cleanly, with full handover documentation and pairing time with your engineers.

Can our security team review the work?

Yes, and we recommend it from the first briefing. Our security engineering team walks your reviewers through architecture, data flows and controls directly.

Tell us about the workflow.

We reply to every enterprise enquiry within one business day. Bring your security team; we will walk them through the architecture and every control.

Arrange a briefing Our security practice