Thoughts

Organisational Memory Is the New Moat Nobody Meant to Build

The moment an AI starts learning your internal workflows, your vendor relationship quietly becomes a hostage situation.

Every enterprise AI play right now is dressed up as productivity. More throughput, less toil, faster decisions. That framing is mostly true and almost entirely beside the point. What’s actually happening is a land grab for organisational memory, and most companies haven’t noticed they’re the land.

Context Is the Product

The useful thing an AI embedded in your communication tools learns is not how to answer questions. It’s the texture of how your company actually works. Which decisions get made informally. Which teams talk to each other and which don’t. What the real priorities are versus the stated ones. That kind of context takes years to accumulate in a human team. An always-on AI agent can hoover it up in weeks.

Once a model has that context baked in, switching becomes genuinely painful. Not because the software is hard to migrate. Because the institutional knowledge it has absorbed doesn’t export cleanly to a competitor’s system. You’re not locked in by contracts. You’re locked in by your own organisational history.

The Swap Cost Nobody Talks About

We’ve had this conversation before with CRMs and project management tools. Data portability became a political issue because everyone realised too late that their workflows were encoded in the vendor’s schema. AI agents embedded in daily operations are the same problem, except the knowledge is fuzzier, harder to audit, and woven into model weights rather than database rows.

The companies building these tools know this. Pitching “your AI learns your company” is a feature. The lock-in is a business model. That’s not cynicism, it’s just product strategy.

What organisations should be asking right now is not whether the tool is useful. It usually is. The question is what they get to keep if they ever want to leave.

That question is worth asking before the AI knows more about your company than your own leadership team does.

Related