There’s a slide near the front of most strategy decks now. “AI-First Organisation.” Bullet points about agentic workflows, augmented decision-making, intelligent automation across the value chain. The board nods. Everyone in the room agrees this is the direction.
Then the meeting ends and nothing about how a decision actually gets made changes.
Same approval chain that took three weeks to route a vendor contract last year. Same four-level escalation for a data-access request — the agent just waits at the chokepoint a person used to wait at. Same architecture review, arriving after the pilot’s already committed. The latency hasn’t gone anywhere. It’s just been automated — which looks like speed until you notice nothing that used to take four levels of sign-off takes fewer than four now.
The strategy moved in one meeting. The machinery that turns a strategy into a decision didn’t move at all.
This isn’t a rollout problem. Rollout assumes the plumbing underneath is ready to carry something new and just needs the thing switched on. The plumbing was built for a different cadence, a different risk tolerance, a different-sized decision entirely. Rebuilding it is slow, unglamorous, and doesn’t photograph well for a board pack. Approving a slide does. So the slide gets approved and the plumbing gets left alone, and six months later someone’s asking why the “AI-first” initiative feels exactly as slow as everything else did.
An operating model doesn’t become AI-ready because a strategy says AI-first. It becomes AI-ready when the thing that routes a decision can carry the speed, volume, and judgement AI produces. Most organisations are running that second thing on infrastructure sized for the first one, and calling the gap change management.
The slide changed fast. Nobody’s put a timeline on the part underneath it.
— Signal / unfiltered observations from the field. Fortnightly.
#EnterpriseArchitecture #AIStrategy #OperatingModel #DecisionGovernance #OrdoAnimi