INDEPENDENT EXPLORATION
Interesting things
happen between.
Experiments at the edge of intelligence,
networks, and simulation.
A small lab for connecting models, agents, data, and worlds—and seeing what emerges.
Explore the experiments01 / AREAS OF EXPLORATION
In the lab.
Early ideas, open questions, and systems taking shape.
01Intelligence
MODELS ↔ AGENTS ↔ TOOLSWhat can a model do when it can work with the world around it?
Exploration
Intelligence
MODELS ↔ AGENTS ↔ TOOLSLocal inference, agent workflows, and useful conversations between machines. Exploring how models can reason, use tools, and carry a task across different systems.
What belongs on your own hardware? Where does an agent need help? How do separate tools become a coherent workspace?
chat.pi0.com02Prediction
SIGNALS ↔ MODELS ↔ OUTCOMESFinding the distance between a forecast and what actually happens.
Concept
Prediction
SIGNALS ↔ MODELS ↔ OUTCOMESWeather, uncertainty, and the signals hidden in changing conditions. An exploration of forecasts that can be compared, measured, and improved.
When do models disagree? How well calibrated are their probabilities? What can we learn by following a prediction all the way to its outcome?
weather.pi0.com03Worlds
RULES ↔ ECONOMIES ↔ EMERGENCESmall rules. Persistent worlds. Consequences nobody scripted.
Exploration
Worlds
RULES ↔ ECONOMIES ↔ EMERGENCESimulated worlds where resources, movement, and human choices shape an economy. Start with the relationships that make a place interesting, then build the world around them.
What makes a route worth traveling? How do scarcity and risk change behavior? Which simple rules make a world feel alive?
world.pi0.com04Systems
MACHINES ↔ NETWORKS ↔ SERVICESThe quiet infrastructure that lets independent things work together.
Exploration
Systems
MACHINES ↔ NETWORKS ↔ SERVICESConnections between local compute, remote services, and the interfaces that make them usable. The underlying experiments that give the rest of the lab somewhere to run.
What should run locally, and what belongs at the edge? How do we make connected systems understandable, observable, and easy to change?
api.pi0.com · status.pi0.comThese are areas of exploration, not public product releases.
02 / A NOTE ON THE NAME
Connected components.
Unbounded possibilities.
In algebraic topology, π₀ names the set of path-connected components of a space: the pieces you can move within, but not between.
For this lab, it’s a starting point. Intelligence, networks, prediction, and simulation each have their own language. The interesting work happens where those worlds meet.
Find the separate pieces. Explore the connections.