AI agents are a boardroom conversation, not an IT project
Why the companies getting real value from AI agents start with strategy and governance — not a proof of concept that never leaves the lab.
Every enterprise I speak with right now is running an AI pilot. Far fewer can tell me what business outcome that pilot is supposed to change. That gap — between activity and advantage — is the single biggest reason AI agents stall after the demo.
The pattern I keep seeing
A team builds an impressive agent. It summarises documents, drafts emails, answers questions. Everyone in the room nods. Then it meets reality: it has no access to the systems of record, no owner, no guardrails, and no measure of success. Six months later it’s quietly switched off.
The problem was never the model. It was that an organisational change was treated as a technology experiment.
Start with three questions
Before a single agent is deployed, I push leadership to answer three things:
- Which decision or process are we trying to change, and what does “better” look like in numbers?
- Who owns this agent in production — and what is it allowed and not allowed to do?
- How does it reach the data and systems it needs without creating a new silo or a new risk?
If you can answer those, the build is the easy part. If you can’t, no amount of engineering will save the initiative.
Governance is the accelerator, not the brake
The instinct is to treat governance as the thing that slows AI down. In practice it’s the opposite. Clear ownership, access boundaries and success metrics are what let you scale from one agent to twenty without losing control — exactly the same lesson the integration world learned with APIs a decade ago.
That’s the work I help leadership teams do: turn the excitement around AI into a portfolio of agents that are measured, owned and safe to grow.
Want to talk through where agents could create leverage in your organisation? Book a conversation — the first one is on me.