Context and continuity
How to preserve decisions and organise long-running work with AI.
Notes
Ideas for building and directing AI systems.
Architecture, context, evaluation, capability transfer and governance. Technical work explained through the decisions it helps you make.
Technical practice
Sources and evaluation
How to check information and design a verification process.
Instructions and controls
Where rules belong, which mechanisms enforce them and how to test them.
Capability transfer
Learning on the real case so people can direct the system.
Full archive
Engineering, decisions and European context.
What your AI knows today and will have forgotten tomorrow
Context engineering decides the quality of the answer: models degrade measurably as the conversation grows, and continuity of work with AI is what makes it avoidable.
Read →Training your people in AI: a driving licence, not a course
AI training for employees that leaves capability, not diplomas — and covers, along the way, the AI literacy that Article 4 has required since February 2025.
Read →What your AI is forbidden to say: compliance that runs with the system
AI Act compliance can live in a PDF or run inside the system. The difference shows on the day of the error. How a legal boundary that executes is built.
Read →How to know your AI is not making things up: source verification
Artificial intelligence hallucinations are not solved by trusting more: they are solved by demanding the source. How verification is built, with a real system open to inspection.
Read →The rules your AI obeys: where they live and how they are enforced
Controlling artificial intelligence is not watching it: it is having its rules live in writing, apply to every response, and be provably enforced.
Read →A Monday morning with AI running the department
Operating with artificial intelligence shows in the routine, not the demo: what the system does by Monday morning and how the manager checks the log.
Read →From using AI to directing it: control as a management capability
The reason to govern your AI is not the auditor: it is directing artificial intelligence as you direct a team. Compliance is the consequence, not the motive.
Read →AESIA: what it is and what it asks of your company
AESIA is the Spanish Agency for the Supervision of Artificial Intelligence: what it is, what it can ask of your company, its sandbox and its fifteen compliance guides.
Read →How to comply with the AI Act, step by step (without becoming an expert)
A roadmap to comply with the European AI Act without being an expert: what AI you have, what system the law asks for, in what order and with which official guides.
Read →Your board cannot see what your AI does: how to change that
Your board signs off on an AI it cannot see without an interpreter. What AI transparency for decision-makers means —human oversight and traceability— and how to achieve it.
Read →Fines under the AI Act: how much, on whom and from when
The European AI Act sets fines of up to 35 million euros or 7% of worldwide turnover. What non-compliance with high-risk AI costs, who is exposed and from what date, in plain language.
Read →No data governance, no AI: why projects stay a demo
An AI is only as good as the data it can read. Why the order of your data —not the model— decides whether an AI project ever operates the business. In plain language.
Read →Why AI pilots fail (and what separates them from the ones that reach production)
Why an AI demo can stall before production: data, integrations, evaluation and operational responsibility.
Read →AI governance for decision-makers, without being technical
Before auditing anything, decision-makers must be able to see what their AI does, what risks it carries, and how it is sustained. Explained without jargon.
Read →Is your AI "high-risk"? How to know without being a lawyer
What high-risk AI is under the EU AI Act and how to know whether yours qualifies: the Annex III list with examples and the documentation catch, in plain language.
Read →Start with the problem you want to solve.
Tell us what you need to build or improve. Technical work starts with understanding your operations.