Home / How We Work

// How we work

Small, fixed-scope steps. Clear go/no-go.

Start with an assessment or go straight to a pilot. Every stage has acceptance criteria agreed up front and ends at a gate, so you can stop with a written recommendation in hand.

01

Assessment

About 2 weeks

02

Pilot

About 4–6 weeks

03

Production

Milestones

04

Run or hand off

Your cadence

01

Agent Readiness Assessment

Map one or two workflows, the data boundary, and the systems. Rank agent candidates by value and risk.

Gate: pick one agent, or stop with the roadmap.

02

Pilot Sprint

One agent, one workflow, sample or redacted data. Acceptance criteria are written before code. Cost per run is measured. The fee is in the proposal.

Gate: evals pass for the agreed period.

03

Production Build

Integrate, harden, and deploy on AWS, Azure, or GCP with observability and a rollback path.

Gate: go/no-go signed.

04

Run, improve, hand off

Keep evals and upgrades moving, or train your team and return the keys.

Exit: the team can run it without us.

Shipped with every agent.

The same controls we teach in the production courses.

  • Step limits
  • Human approval on writes
  • Eval gates
  • Cost budgets
  • Kill switch

How an engagement is run.

  1. 1NDA

    Before any details

  2. 2Proposal

    Within 24 hours

  3. 3Founder-led

    We stay on it

  4. 4Your IP

    The work stays yours

  5. 5Any gate

    You can stop

  6. 6No secrets

    Not in the intake form

Can we skip the assessment?

Yes, when the workflow and the risk are already clear. The proposal says which gate we start at.

What data do you need?

Sample or redacted data for the pilot. Production access comes later, under your controls.

Who owns the IP?

You do. The handoff includes the repo, the evals, and the runbook.

Can you train the team instead?

Yes. The Agentic Production Series is the public version. Private cohorts are scoped on request.

Request a proposal.

Request a proposal →