Strange Lab

We're building world models for work.

Frontier LLMs can reason and use tools once work is specified. Open-ended work starts earlier: choose what matters from thousands of possible actions.

A world model learns how a repository, trial registry or company changes from its dated history. We turn those models into tools for forecasts, triage and conditional what-ifs. An LLM reads the evidence, chooses an action and uses tools; we evaluate the decisions and their later outcomes.

We’ve trained world models across software, clinical research, agents and companies, and tested them against later outcomes.

The same tools give people and agents a way to choose where to focus, inspect the evidence and compare possible next steps. Research shows the results and working examples.

We start with one bounded world, test transfer across related worlds, then build domain and industry models. Our world models essay explains why the same approach may help agents work in open-ended environments.

See how raw data becomes a world model, explore an event stream, or play against a poker agent.

Explore Company Lab · By invitation. Follow company records, forecasts, plans and a recorded operating year.

Our goal: simulate every decision, across the economy.