Twenty years of architecture and operations inside large corporations and major IT services firms. I help IT leaders from Lyon to Geneva decide, frame and govern what they hand over to AI, without giving up the keys to the machine room.
Damien Cortès — enterprise architect, adviser to CIOs
Français · English · Deutsch
Every vendor now has its own AI agent, and each one promises to soak up your incidents. Between the demo and the 3 a.m. call-out lies a gap. My job is to measure it, close it where that makes sense, and say so plainly when it doesn't.
What in your operations can genuinely be handed to AI, what can be assisted, and what must stay human. And, among the agents you are being pitched, which one to pick — or none at all.
“ITIL isn't dead. It just has to learn to work with agents.”
Incidents, changes, problems: a ladder of autonomy, one rung at a time, with a rollback at every step.
What must stay in-house will stay in-house: open models on-premises or in a Swiss or French cloud. And teams who can read, challenge and sign off what the machine suggests.
Consultant, founder, architect: twenty years between the machine room and the boardroom. That is exactly why I now work outside it.
A master's in networking at Lyon 1, a degree in information systems in Karlsruhe, and EM Lyon's specialised master's in entrepreneurship. First freelance assignments: operations for a great name in luxury goods in Épernay, then a German development team in Frankfurt.
Code and business, from day one.
Data centre and operations consultant at Accenture: SAP infrastructure for the public sector, mainframe optimisation for a British bank, IT service management for an airline.
Learning operations where mistakes are not forgiven.
Founded in Lyon in November 2010. Six years of assignments with major accounts in the region and beyond:
Independent, and already at the heart of major accounts.
Pre-sales and infrastructure architecture for bids worth over €10m (energy, aerospace, defence, banking, chemicals, luxury, agribusiness), then solutions architect on large cloud and infrastructure contracts, with teams spread from India to the United States. And enterprise AI: moving from experiments to the structured adoption of large language models.
Large corporations, seen from the inside.
A personal infrastructure run like a production estate: AI agents under human approval, local models, monitoring, automated releases, backups restored and tested every week.
Try it on myself before advising anyone else.
Cortès Consulting sets sail again, from 39 rue Barrier in Lyon, facing the Lyon region and the Léman area: Geneva, Nyon, Lausanne.
Independent once more, with twenty more years and machines that learn.
Agents that take a share of the night shifts, data that stays with you, distributed ledgers where trust has to be proven, and teams that keep their hands on the controls.
The future of operations is being built now, and slowly.
Before I mention anything to a client, I run it at home, day and night, to production standards: monitoring, logs, backups, rollback.
They propose, a human approves. Every action is logged.
Open models, with no data leaving the building.
Monitoring, automated deployments, tested backups.
A node in service, to understand the technology before the promises.
agent › incident: disk at 91% agent › probable cause: logs not purged for 9 days agent › proposal: purge > 30 d, estimated gain 42 GB waiting for human approval… approved by the operator agent › done, disk at 64% agent › rollback: backup from 03:40, tested Monday ✓ closed in 6 min, logged
These are not promises with dates attached. They are the subjects I am working on, and I would be glad to discuss them with you.
In a few weeks, an honest measure of what can be handed to AI, assisted, or must stay human. A roadmap, not a brochure.
Taking an agent up the ranks, from read-only to supervised action. Every rung is earned through measurement.
What they really do, where your data goes, how you switch them off. No vendor commission.
Open models on-premises or in a Swiss or French cloud, and a frank measure of what local models can actually do.
A workshop where teams replay real incidents with an agent at their side, and learn to push back on it.
Incidents, changes, problems: the processes you know, brought up to date for the decade ahead.
AI Act, Swiss and French data protection, regulated industries: translated into concrete operating rules.
Licences, compute, human time: what an incident resolved with AI really costs, before you sign.
Bitcoin, Lightning and enterprise blockchains: where a shared proof genuinely adds something (traceability, timestamping, settlement), and where a database will do.
Preparing for certification where it is won or lost: in production. Evidence collected continuously, automated controls, and AI agents that document what they do, so the audit is no longer a scramble.
Meetings between heads of operations from Lyon and the Léman area, away from trade fairs and sales pitches.
A first one-hour conversation, with no strings attached: you describe your operations and what you are being offered, and I tell you frankly what I think.