I lead digital experience inside a large industrial group. This blog is where I think in public about the technological and anthropological shifts running through work today—inside organisations, and beyond them.
“The future always comes too fast and in the wrong order.”
— Alvin Toffler, Future Shock, 1970
Caravels and Corsairs
In the last days of October, fleets of AI agents came back with claims on theorems and with formalized proofs. The history of the great explorations suggests what comes next: charts, trade, corsairs, and inhabited continents.
The Enterprise Street View
Walking through an enterprise, you see walls and closed doors. An AI agent turns the walk into a map: it can survey the whole urban plan of the company, and it needs to bypass no access control to do so.
From Patches to Proofs
Crisis after crisis, a digital empire too vast to defend by patches alone. Anthropic's new work on formal verification suggests that the lasting answer may be mathematics: software that carries proofs of its security properties.
Firebreaks
In a few days of October, shared code, a shared identity and shared infrastructure failed. Three mechanisms, one shared question: how far does a failure travel, and who is responsible for planning the digital city?
Repair
A mountain in Cambodia, scarred by history and by scam compounds, and a small church that still stands. A question for everyone who builds, funds, deploys or governs digital technologies.
Chevaux de bataille
Au Mondial de l'Auto, Citroën dévoile une 2CV électrique qui vise un prix proche de 15 000 euros. Sur le flanc du pneu, le nom « Michelin » se lit comme en hommage appuyé à l'héritage qui jadis unit les deux industriels.
Les Phares
Pour le mathématicien Hugo Duminil-Copin, les grandes conjectures sont des phares dans la nuit. Si l'IA les résout du jour au lendemain, on gagne des réponses et on perd peut-être ce que leur recherche faisait naître.
Free Rein
The time to experiment with AI is a management decision before it is an individual one. The word manage itself began with handling a horse by hand.
Les 9 vies d'Aston Martin
91% de capitalisation boursière évaporée en huit ans, sept faillites en cent treize ans—le constat de Thierry Picard est exact. Mais Aston Martin n'est peut-être pas une entreprise qui possède une marque. C'est un mythe auquel, depuis 1913, on essaie de donner une entreprise.
A Few Blocks Apart
Notion has spent months opening its blocks outward to agents, OpenAI just built its agents a workspace to live in, and Anthropic folded Docs and Slides into Claude. Three companies, three different starting points, converging on the same shared object within weeks of each other.
Syllogism?
A new architecture lets a model rewrite its own context, arbitrarily, in place. The syllogism can stay perfectly valid while the premises quietly narrow underneath it.
Control Plane
In 1936, an engineer counted airplanes and found a curve that seems to price AI tokens today. What governs technology, the control plane, was built for a slower pace and may need a redesign of its own.
Think (Not So) Different
What laptop should a company buy its people for the next four years? Microsoft and Nvidia share a stage this week to answer that question—but most of the answer may have already quietly shipped since June, while agentic AI might have weakened the foundations of Microsoft's historic rent.
Gradient Descent
A paper on the economics of information, a theory of disruptive innovation and a bet the frontier AI labs are about to make all describe the same thing: value lives in a gradient, and using it is what spends it.
The Thick Cloud
Copilot is now pitched as 'a new OS for work,' while a wave of small, capable models is turning ordinary laptops into local inference nodes. After a decade pushing work into an ethereal cloud, the sky seems to be clouding back over.
Decision-as-a-Service
A new category of model doesn't generate text. It decides—fast, cheap, and typed. What it recovers may matter more than what it adds.
Track Record
To explain the 2026 AI investment supercycle, Satya Nadella pointed the industry toward 1873, not 1969: the year railroad financing, much of it circular, collapsed under its own weight. The panic destroyed the capital. It never touched the railroad—the distinction this AI cycle keeps collapsing into one claim.
New pieces, infrequently. Subscribe.