Knowledge Flywheel: The memory makes the difference

In two years, many professional services firms will likely be running the same AI assistant. On the same frontier models, with comparable capabilities.

So what sets the firms apart? The memory they bring to that AI.

Almost no one has structured that memory. It lives in the heads of a handful of senior partners: client knowledge, negotiation history, methodology DNA — knowing that a particular client consistently holds a hard line on a specific clause lives nowhere systematically, only in memory. Each departure takes a piece with it.

Pull it off, and you have something competitors can't simply buy. Don't, and in two years what you have is a faster search engine.

I have tried to lay out an architecture for this. What sits inside the firm has to be separated into three layers. Who curates what, and who approves it, has to be named. And the vocabulary has to grow with reality, otherwise within two years the AI is working against a world that no longer exists. Deliberately no tool recommendation — that layer ages faster than the architecture beneath it.

https://sixtyfour.solutions/praxis/wissens-flywheel/

If you have hands-on experience with knowledge bases for AI, or you see a missing point: comments are welcome.