Édition 2026 Talk Data & AI Operating Models

AI will make your organization obsolete. If you have one

Langue EN · FR

Speaker

Alexis Blandin

Alexis Blandin

Head of Data @ Jolimoi

Description

Two organizations, one operator, same data leader.

I lead data for both: Jolimoi, a European retail startup, and Urban Sketchers, a global non-profit organization prompting on-location drawing. We run the same agents, same data strategy with the same person rolling them out but with opposite outcomes. One had to be partially rebuilt. The other barely noticed.

The talk argues that agents don't automate work so much as they make an organization obsolete but only the parts that had turned roles into positions instead of responsibilities. Everything else, they leave standing.

Jolimoi was a division-of-labor machine: silos, agile ceremony, ticket queues, territories to defend. The agents dissolved most of it. We shrank the data team without losing ground, ended push-mode for pull, produced far more decisions, and moved trust from humans to machines then learned the hard way what that costs in a non-deterministic world. What looked like automation was really recomposition: the org was becoming something older.

Urban Sketchers was already there. All volunteers, spread across the globe, no scaffolding, no playbook to obey, no title or budget to protect. Every role is already a responsibility. The same agents changed nothing structural but just made the work lighter.

The talk opens and closes in a Renaissance workshop (aka the bottega) the exact opposite of what Adam Smith described: fluid in skills, fixed in responsibility. Urban Sketchers never left it. Jolimoi is being pulled back toward it. That may be where AI-driven organizations end up: an agent can make a skill obsolete, and a position, but never the responsibility underneath.