01 / WATCH
See the work.
Follow decisions, evidence and handoffs. Observe the team, not just the answer.
A research lab you can walk into
Change the rules.
See what happens.
Organizations don't run on intelligence alone. They run on decisions, coordination and ways of working. We're making those mechanics visible — with people, AI agents and everything in between.
CONTROL ROOM / 01WATCH · MIX · REPLAY
aolabSTUDIO
EXP 01 · INCIDENT ROOMOpen the studio
Smart individuals don't automatically make a smart organization. So what does?
That's the experiment.01 / The Studio
Less theory on a slide.
More experiments on a screen.
The Studio lets us observe how teams of agents behave, change their operating rules and inspect the consequences. You can explore the first walkthrough today.
02 / MIX
Change the rules.
Try a solo agent, a central coordinator or a peer-review network. Compare operating principles.
03 / REPLAY
Inspect the result.
Go back through the event trace, challenge the decisions and compare runs.
First experiment / Incident Room 01
Same challenge.
Different organization.
Six fictional incidents. One fixed priority rubric. Different arrangements of agents. What changes when information flows differently?
- Solo generalist, hub-and-spoke or peer network
- Speed-first or evidence-first operating principles
- Recorded decisions and traceable handoffs
Incident Room 01STUDIO DEMONSTRATION
AgentSpecialists
AgentCoordinator
OutputDecisions
3Team structures
6Fixed incidents
2Constitutions
Live model execution is currently controlled-access. The illustrative walkthrough is open to explore.
02 / The research
Three kinds of organization.
One big question.
How do groups become more capable than the individuals within them? We study the mechanisms across three settings — and what carries over between them.
Human
How people share knowledge, make decisions, build trust and develop operating habits that last.
Human × agent
What should agents decide? When should humans step in? Where does accountability sit?
Agentic
Can autonomous agent teams coordinate, learn, recover from mistakes and scale effectively?
Featured research / XC
Can we move
the exponent?
What if each successful transformation made the next one easier, faster and more valuable? XC explores the mechanics of increasing returns from organizational change.
Explore XCAn open hypothesis · XC-01 field protocol drafted · No field results yet
03 / How we work
Good experiments.
Better questions.
Start with a claim. Build a test. Change one thing. Watch what happens. Publish enough of the method that someone else can challenge it.
A convincing simulation isn't the same thing as evidence about people. We'll be explicit about what a result does — and doesn't — establish.
Join the experiment
Have a better question?
We're building AOLab with researchers, organization designers and people who work with agents. If you want to test an idea, compare methods or run a study, get in touch.
aolab@oxford.ai