Views from engineers, not PR.
Analyses from real projects: AI, architecture, automation, product and data sovereignty.
22 articles
EnerJ-meeting Lyon 2026: after the measurement, who decides
Stand ST13bis, Palais de la Bourse. Visitors already had charts. They came to see who detects the drift, and whether the data stays with them.
From AI proof of concept to production: the five recurring blockers
Moving from PoC to production is rarely a model problem. It is an organization problem, and constraints left unnamed too early.
Semantic layer: why your AI agents go astray without one
An agent does not understand your company. It navigates what you made interpretable.
AI agents: human control as a production requirement
What eighteen months of agents in production change about how we design human validation.
Agent governance: logs, scope, human escalation
Without logging and clear escalation, an agent is not governed. It is merely launched.
Evaluating an AI vendor: an operational checklist for CIOs
An AI vendor is judged less by its demo than by what it lets you inspect, export, and stop.
Custom or SaaS: framing the problem correctly
A decision frame from assessments with CIOs and business leaders - without ideological camps.
AI Solution Factory: industrialize AI the way you industrialize software
Industrializing AI is not about faster demos. It is about accepting the constraints of a product that must last.
Enterprise RAG: what actually works beyond the demo
Useful RAG is judged less by fluent answers than by source traceability.
No articles match your search.