Service: AI Engineering Practice
FOUR TO SIX WEEKS · PRACTICE ENGAGEMENT

Claude Code, agentic engineering, AI coding adoption.

Turn AI engineering into a practice your team operates, not a hobby a few engineers run.

Agent workflows, review rituals, evaluation harnesses, an internal playbook. The engagement that takes a team using Claude Code or Cursor and makes the work auditable, repeatable, and worth scaling.

Book a fit call
Run an AI engineering audit

One audit worksheet, no follow-up sequence. We send it once.

Starts from EUR 10k. Four to six weeks, plus an optional retainer.

01
WHO THIS IS FOR

You are past the hype curve and into the work.

  • 01 Your engineers are already using Claude Code, Cursor, or another agent in their daily work.
  • 02 Output has gone up, but you cannot point to a repeatable practice.
  • 03 The good results come from two or three people. The rest of the team is watching.
  • 04 A review process exists for code, but not for what AI generated, when, and against what intent.
  • 05 You want a discipline, not a vendor pitch.

If three of five ring true, the practice is missing, and the practice is what compounds.

02
WHAT YOU GET

An operating practice, not a training course.

Part 01

Agent workflows

The specific, named loops your team will run: code generation, refactor, review, test, document. Each loop has an owner, an input contract, and an exit criterion. No vibes-based agent work.

Part 02

Review rituals

A rubric for reviewing AI-generated work: what the human checks, what the runtime enforces, what the team escalates. The reviewer is not optional. The rubric makes the review fast.

Part 03

Evaluation harnesses

The lightweight test suite that tells you whether an agent is regressing. We do not build a model evaluation lab; we build the minimum your team will actually run on Monday.

Part 04

The internal playbook

A written document, owned by your engineering lead, that captures the practice. Onboarding new engineers takes hours, not weeks. Anybody can read it.

03
HOW THE ENGAGEMENT RUNS

Five weeks, embedded.

  1. Week 1

    Audit.

    With your team

    We watch the team work. We name what is already a practice (even if it has not been written down) and what is improvisation. We pick the three workflows worth codifying.

  2. Week 2

    Codify.

    With your team

    We write the workflows with the team. Owners, input contracts, exit criteria. The rituals become explicit.

  3. Week 3

    Review and evaluate.

    With your team

    We design the review rubric and the evaluation harness. They have to be small enough for the team to actually run.

  4. Week 4

    Embed.

    With your team

    We run the new practice with the team for a week. We tune. We document.

  5. Week 5

    Hand off.

    With your team

    The internal playbook lands with the engineering lead. The retainer (optional) is monthly office hours: review the practice every four weeks, adjust as the tools and the team change.

04
THE GRADUATION PATH

Where the practice lives next.

You leave with the workflows, the rituals, the evaluation harness, and the playbook. They are yours; if you never work with us again, the practice survives the next tool cycle.

If we do continue, the practice runs against Process Native Software. The runtime carries governance and audit so AI-generated work is reviewable by default, not by exception. A change generated by an agent is a change governed by the runtime. The reviewer is faster because the floor is doing half the work.

05
THE MAP
A practice is what survives the next vendor migration. A toolkit is not.

EmpoweredHouse · AI Engineering Practice

06
PRICING

Starts from EUR 10k.

What changes the number: the team size, the surface area (one repo or many), whether compliance review is on the critical path. Retainer is priced separately; most teams take it for the first quarter, then drop it.

Team size

How many engineers the practice must reach.

Surface area

One repo or many.

Compliance review

Whether compliance is on the critical path.

We will quote the engagement in the fit call. If the team does not have a working practice yet, this is the wrong engagement, and we will say so.

07
THE FIT CALL · 30 MIN

Find out whether the practice is what you are missing.

A fit call is thirty minutes. If your team is already shipping with AI, we will ask how, why, and what is repeatable. By the end, you will know whether the practice is what you need, or whether you need something earlier in the stack first.

Book a fit call Run an AI engineering audit