Arcana Advisory
Book a time

AI implementation and engineering

Forward-deployed engineers ship what you decided into production fast, under rules your team can read.

Architect 10Deliver 20Run AI-assisted engineering 3040 Govern50 Hand overTECHNOLOGYREV 2026-11 · SHEET 2/5FIG. 1SHEET 2 OF 5FIRETECHNOLOGY10Architect20Deliver30Run AI-assisted engineering40Govern50Hand overREV 2026-11 · SHEET 2/5FIG. 1SHEET 2 OF 5FIRETECHNOLOGY
Practice
Technology
Led by
Dominique Legault

What does an AI implementation partner do?

An AI implementation partner takes a decision and gets it built. Our technology practice covers architecture, delivery and AI-assisted engineering, from choosing what to build to running it in production, including the governance that keeps AI coding agents accountable. We work embedded with your team or on a regular rhythm.

What we do

  1. Architect

    Decide the shape of the system, the build-or-buy calls and the order to build it in before code is written.

  2. Deliver

    Work alongside your engineers on the core of the system until it runs in production.

  3. Run AI-assisted engineering

    Set up how your team uses AI coding agents: what they may do alone, what a person must approve, and how the work is reviewed.

  4. Govern

    Put decisions, quality gates and an audit trail around AI-driven development. AI-SDLC, our open-source framework, is built for this.

  5. Hand over

    Leave the system documented and your team able to extend it.

If this is the problem on your desk, book half an hour with the person who leads it.

Book a time

Who leads it

Related Lab work

  • AI-SDLCLive · 2026-11

    The decision engine for autonomous software development

Questions people ask

Do you build things or only advise?

We do both. The technology practice works hands-on with your engineers to design and ship, and it also advises on build-versus-buy and delivery. What we do on your project is agreed on the first call and written down before work begins.

How do you work with AI coding agents safely?

By deciding in advance what an agent may do alone, what needs a person to approve, and how its work is reviewed. AI-SDLC, our open-source framework, turns those decisions into a spec, quality gates and a record, so review happens on the plan and not only the diff.

Can you fix an AI project that stalled at pilot?

Often, yes. The usual causes are an unclear decision, no owner, or a build that skipped security and data questions. We start by naming which of those it is, then say whether the project should be finished, reshaped or stopped.

Will you work inside our stack and tools?

Yes. We work with the languages, cloud and tools you already run. AI-SDLC is agent-agnostic and connects to several AI coding tools, so the practice does not depend on one vendor.

Is AI-SDLC something we have to buy?

AI-SDLC is open source and public. You can read it, run it and judge it before you talk to us. Link and details are on the AI-SDLC page in the Lab.

More questions, answered in the Grimoire.