AltFn
Contact us
BlogAltFn

World Models for Organizations.

We study how to learn the state of an organization from the traces of its people and agents, how to rehearse changes against it honestly, and how agents earn the right to act in it.

Start with the memo

Posts

Routes, Not Retrieval

Enterprise agents can already find the facts. What they're missing is the route: who signs off, in what order, and what tends to go wrong.

Read the post

The Self-Driving Enterprise

Every team is about to work alongside AI agents. Leaders need line of sight into how the organization actually works. The agents need what every robot needs: a model of the world they work in.

Read the memo

Research

Q1

Can an organization's next state be predicted in latent space from its own traces?

Q2

How sure does a prediction need to be before an agent acts on it?

Q3

When has an agent earned the next level of autonomy, and who decides?

Q4

Can a model of the whole organization respect every source system's access rules?

Working on world models, RL or evaluation and want to collaborate? research@altfn.ai