Insights and
field reports
Analyses, field reports, and insights on AI, tools, and best practices.
Latest articles
When Should AI Queries Run Locally?
How to measure which AI queries can run locally: routing architecture, cost, latency and confidentiality.
Which AI Model for Your RAM: A Guide
Choosing an open source AI model by your available memory: what runs on 8, 16, 32 or 48 GB, quantization made simple, and the sovereign workstation.
Why Your AI Agent Never Learns
An agent with no memory is as ineffective on its hundredth use as its first. The 3 types of memory and the writing discipline, for a business.
AI Agents & Claude Code
Building, deploying and maintaining AI agents in production. Claude Code, MCP, Skills, Hooks, multi-agent orchestration.
Why Your AI Agent Never Learns
An agent with no memory is as ineffective on its hundredth use as its first. The 3 types of memory and the writing discipline, for a business.
Performance Comes From the Harness
Two companies, the same AI model, opposite results. The difference is not the model but the harness. The 5 layers explained for a CTO or CISO.
The CIO Guide to Deploying Claude
Before rolling Claude out to a team, the calls to make: which plan, which integrations, where inference runs, and how to govern team access.
Claude Certified Architect: the official certification
Anthropic launched Claude Certified Architect (Foundations), the first official certification for architects building AI agents. What it changes.
Why Your AI PoC Isn't Production-Ready
Three things separate an AI proof of concept from a reliable production agent: evaluation datasets, guardrails and observability.
Claude Code Hooks: Automate Quality Without Thinking About It
Hooks intercept every Claude Code action to lint, test, and secure automatically. 5 concrete hooks you can copy-paste into your project.
AI Security
Securing AI agent deployments at scale. OSINT, supply chain, guardrails, audit.
Your AI Instruction Files Are Code
A CLAUDE.md or agent config downloaded from the internet is executable code. The real vectors of the AI supply chain and a 6-point checklist.
What Attackers See About You
Passive reconnaissance reads what your company exposes without touching your systems. What is visible, why AI helps attackers, and 5 checks to run.
Securing AI Agents in the Enterprise
How to secure AI agents that act on your systems: key trade-offs, seven defense layers and common traps for CIOs and CISOs.
AI Act & Compliance
The EU AI Act: obligations, timeline, scope. What actually applies to your organisation.
EU AI Act: What Changed on 2 August 2026
The EU AI Act now generally applies. See what changed in August 2026, what the AI Omnibus postponed, and which actions remain urgent.
EU AI Act for SMEs: What Applies From August 2026
A practical EU AI Act playbook for SMEs: classify each use, identify your role, collect evidence and build a proportionate compliance plan.
Sovereign & Local AI
Your data stays with you. Ollama, local Mistral, choosing a model by RAM and sensitivity.
When Should AI Queries Run Locally?
How to measure which AI queries can run locally: routing architecture, cost, latency and confidentiality.
Which AI Model for Your RAM: A Guide
Choosing an open source AI model by your available memory: what runs on 8, 16, 32 or 48 GB, quantization made simple, and the sovereign workstation.
Sovereign AI: your documents never leave
Where do your documents go when you use a cloud AI? For a lawyer, an accountant or a clinician, the answer should be: nowhere. Here is why and how.
Confidential AI: Data Stays On-Premises
How to use AI without sending your data to Google, OpenAI or Microsoft. Installed on your machines, free models, zero subscription. We explain everything.
Training & Qualiopi
Funding AI training programmes. Pricing, French OPCO, Qualiopi certification.
Corporate AI training cost: 2026 price ranges
How much does corporate AI training cost? 2026 market ranges, what drives the price, net cost after French OPCO funding, and real budget examples.
OPCO Funding for AI Training
How to fund AI training through an OPCO in 2026: Qualiopi requirement, finding your operator, coverage, four steps and the FIF-PL case.
Method & tools
Choosing an AI tool, writing better prompts, encapsulating business expertise. Field reports.
Turning Business Expertise into AI Skills
Your experts' know-how can become a reusable module your AI agents call on: an asset your company owns, not a subscription you rent every month.
Claude Skills Tutorial: Build a Reusable SKILL.md
A practical Claude Skills tutorial: create a SKILL.md, structure reusable instructions, load references progressively, and share the workflow with your team.
Autoresearch: when an AI agent runs ML research for you
Andrej Karpathy's autoresearch lets an AI agent modify training code, run short experiments and iterate autonomously. Here's why it matters.
Prompt Engineering for Business: 7 Patterns + QA Checklist
Seven reusable prompt patterns for business teams, plus a QA scorecard for testing accuracy, evidence, format and safe escalation.
No-Code AI Automation: Where to Start in 2026
Make, n8n, Zapier AI, Claude Cowork. No-code tools to automate your workflows with AI. A practical guide for SMBs.
How to Choose the Right AI Tool in 2026: A Use-Case Guide
ChatGPT, Claude, Gemini, Mistral, Cursor, HeyGen, Make. Which AI tool for which job? A practical comparison by category to make the right choice.
Got a project in mind?
Describe your situation. The first 30-minute call is free. Get a frank response within 48 hours.