Rust CLI
RTK
Reduce the noise in command output.
A Rust CLI proxy for more compact agent context. First and main contributor in Q1 2026, then an evangelist.
Fractional Chief AI Officer
Transform how people and organisations work with AI.
From choosing how to use AI to deploying agents your teams can run and oversee.
Expert in harness engineering
I help your leadership team decide which tasks to give AI agents. I show your teams how to build and configure the right harness for their needs and how to check the agents’ work. I train them using examples from their own work and guide them as they take over. I step in when needed, including to write code.
01 / Services
AI ADOPTION / HANDS-ON LEADERSHIP
From support to engineering, we decide who does what, which steps agents can handle and how teams check the results.
Explore the servicesA RECENT ENGAGEMENT / ANONYMISED
In an ongoing engagement, I trained teams, configured their harness, created a dedicated repository and built a shared knowledge base.
Let's discuss your situation →02 / Learn
Start with the question you need to answer.
Subscribe to the newsletter01 / GUIDE
Start with a setup you understand, then add capabilities deliberately.
Claude Code setup, level by level
02 / GUIDE
Build the context your agents need, from project instructions to shared knowledge.
Context engineering: the L0-to-L5 playbook
03 / ARTICLE
Measure your own usage before choosing tools, subscriptions or limits.
Measuring your own AI usage: three meters, three different answers
04 / ARTICLE
Turn individual habits into shared practices. Start here for the team layer of the context series.
5/6 · The AI instruction system is a product, not a config file
Prefer a sequence? Follow a reading path →
03 / Open source
Rust CLI
Reduce the noise in command output.
A Rust CLI proxy for more compact agent context. First and main contributor in Q1 2026, then an evangelist.
TypeScript Context engineering
Keep agent instructions grounded.
Detect stale facts and check the context your agents rely on in CLAUDE.md, AGENTS.md and related files.
Python MCP
Connect agents to real SEO work.
Structured access to search analytics, URL inspection and technical SEO tools through MCP.
Model, effort and skill routing for coding tasks. Public source; no measured cost gain claimed.
Under the hood / Harness engineering
Follow a request through the context, tools and checks around an AI model.
Explore the workshopI’m currently exploring inference: where to run models, which hardware to use and what it costs. I collect my research and reference material in the Claude Code Guide.
04 / Field notes
ARTICLE ·
A test that passes for the wrong reason leads to a practical way to assess how a team scopes, verifies, explains, fixes and hands over an agent’s work.
ARTICLE ·
Sur 264 identifiants synthétiques, une refonte SQL ramène une requête d’accès de 530 à 3 jointures. Un test d’expiration protège la règle métier.
ARTICLE ·
With Thomas Pierrain and Vanessa Gossent, I revisit AI costs, hiring, parallel agents and the work we have to redo after delivery.
ARTICLE ·
My conversation with Bruno Soulez on context engineering, agent harnesses and code review: what I delegate and what I still check.
ARTICLE ·
Three months, 1,011 pull requests and 8,728 inline AI review comments: 4.5% got a visible human response. What the data shows, and what it cannot.
ARTICLE ·
AI can shorten implementation while specification, review, CI, and rework absorb the saved time. Instrument the full flow before adding more generation.
05 / Speaking
I also speak in small groups with teams and their leaders.