We audit, remediate,
coach and train.
Code & architecture audit
Before deciding on a rewrite, a refactoring or a freeze, know what you have.
We analyze the codebase and its architecture: module boundaries, dependency cycles, coupling hotspots, drift between domain concepts and code. We use our own tooling (Codegraph) and deliver evidence: graphs, metrics, a ranked list of actions.
An audit is also how a remediation starts.
Read more →The real structure of the system, not the org chart’s version.
Where change is expensive today, with numbers.
What to fix first, and what it buys you.
Legacy remediation
Your system works, but every change costs more than the last. We take it back under control.
Our approach, Legacy Reclaiming, combines Michael Feathers’ legacy-code techniques, legacy displacement patterns and verification by behavioral equivalence:
The outcome: a modular codebase your team can change again. The harness stays yours, a permanent non-regression net.
Read more →We characterize the real behavior of the system first, then rebuild next to it. The old code stays the reference until the new one has proven equivalent.
Module by module, no big bang. At no point is production at risk.
Agents execute the mechanical refactorings; nothing lands without passing the harness and a review by our architects. How we do this is public: read our AI Handbook.
Development coaching
We embed in your team and raise the bar on real work: domain modeling, testing strategy, refactoring, working with agentic AI. Pairing and design sessions happen on your production code.
The goal is a team that no longer needs us.
Read more →On your domain, with your constraints.
On your production code.
The workflow from our AI Handbook, in your context.
DDD & craft training
Three courses, Qualiopi certified: Product Design, Domain-Driven Design, Production-Ready Clojure. Jérémie was trained by Eric Evans in 2009 and founded the French DDD community.
Dates, programs and brochures on the training page.
3 days · 18 hours
3 days · 18 hours
4 days · 28 hours