From a toy’s brain to an autonomous combat humanoid under human command.
Learn in simulation. Transfer lessons forward. Validate in the physical world. Only the first step is done.
Where it came from
It started as a small brain for a toy that had to work offline. That is still the rule: it thinks on the machine, not in a data centre. Every step since has asked the same question at a larger scale: how do you build a brain you actually own — one you can open, read and fix?
- Day 1The brain factory
- Feb–Mar 2026TinkyBrains
- Mar–May 2026SML — the smallest language model
- Mar 2026 →Structured knowledge
- Mar 2026 →Trace — Peel
- Sep 2026Fail-First
- Sep 2026Peel — a self-evolving symbolic brain
- Sep 2026One brain, many worlds
- NextThe embodied brain
The road ahead
- 01running
Games and simulation
Doom, Wolfenstein, Fallout, GoldenEye, tanks, a drone flight stack in software, photoreal driving. This is where every number on this page comes from.
crutch it takes awayNothing yet: perfect resets, the game telling it the truth, opponents it already knows.
The game plan ↻
01game planGames and simulation
- The move
- Run the one brain across DOOM, Wolfenstein, Fallout, GoldenEye, tank games, a simulated drone course and photoreal driving. Every attempt and every failure is recorded.
- What changes
- Nothing yet: the game resets perfectly, tells the truth, and the opponents are known.
- The gate
- Learning shown on held-out tests, not just the runs it trained on: drone 10 of 10 clean laps from unseen starts; a tank brain carried into another game at +17%. Losses published alongside.
- What it proves
- The loop works: fail, explain, change one thing, prove it, write the rule.
↺ Back
- 02next
Real controller, simulated world
The same brain on a real flight controller and motor driver; the world is still simulated.
crutch it takes awayPerfect timing: real latency, a real bus, a real clock.
The game plan ↻
02game planReal controller, simulated world
- The move
- Put the same sealed brain on a real flight controller and motor driver, with the world still simulated.
- What changes
- Timing becomes real: real latency, a real bus, a real clock.
- The gate
- The drone-course result holds at real hardware timing, and every floor veto is still on the record.
- What it proves
- The brain survives real electronics before it meets real terrain.
↺ Back
- 03gated
Cheap robots in the dirt
Small ground and air robots, where a crash costs a part, not a program.
crutch it takes awayThe game telling it the truth: the eye has only its own camera.
The game plan ↻
03game planCheap robots in the dirt
- The move
- Small ground and air robots in the dirt, where a crash costs a part, not a program.
- What changes
- The game stops telling it the truth: the eye has only its own camera.
- The gate
- It completes courses on its own sensors, and turns its failures into rules on the robot itself.
- What it proves
- Learning without engine truth, on cheap metal.
↺ Back
- 04gated
Humanoid in our own lab
A commercially available humanoid, on our bench. It never spars against people.
crutch it takes awayCheap consequences: balance, falls and wear are real.
The game plan ↻
04game planHumanoid in our own lab
- The move
- A commercially available humanoid on our own bench. It never spars against people.
- What changes
- Consequences stop being cheap: balance, falls and wear are real.
- The gate
- Stable autonomous routines with a human holding the stop button, and the failure-to-rule loop running on the hardware.
- What it proves
- The same brain can drive a body that can fall over.
↺ Back
- 05gated
The ring: robot vs robot
Sanctioned robot-vs-robot bouts, the remote pilot replaced by the brain. This is where we raise the money and find out what breaks when the other side hits back.
crutch it takes awayKnown opponents: now the other side controls half the experiment.
The game plan ↻
05game planThe ring: robot vs robot
- The move
- Sanctioned robot-vs-robot bouts, the remote pilot replaced by the brain. This is where we raise the money.
- What changes
- Known opponents are gone: the other side controls half the experiment.
- The gate
- Scored on interventions per hour, recovery, new failure types, floor vetoes, rules kept and thrown out, fewer repeat failures, and lessons carried to opponents it has never met. Losses stay public.
- What it proves
- It learns between bouts against opponents we did not build.
↺ Back
- 06gated
Someone else tests it
Someone else runs the protocol, on their robot, against their opponents.
crutch it takes awayUs: the people who wrote the experiment.
The game plan ↻
06game planSomeone else tests it
- The move
- Someone else runs the protocol, on their robot, against their opponents.
- What changes
- We are removed: the people who wrote the experiment no longer run it.
- The gate
- The results reproduce without us in the room.
- What it proves
- It is not our story any more; it is a result.
↺ Back
- 07gated
Combat-capable humanoid, under human command
The goal. A combat-capable humanoid operating under human command: people hold authority over the use of force, the floor holds the machine to the constraints it was given, and every decision is on the record. Not built. Gated on every step above.
crutch it takes awayThe ring: rules, a referee, and a stop button someone else controls.
The game plan ↻
07game planCombat-capable humanoid, under human command
- The move
- A combat-capable humanoid under human command, built on every step above.
- What changes
- The ring’s rules, referee and stop button give way to a commander’s orders and the ROE.
- The gate
- People hold judgment over the use of force, as DoD Directive 3000.09 asks; the floor holds the machine to the machine-readable constraints they set; every decision is on the record.
- What it proves
- Learning can change its tactics. It can never widen its authority.
↺ Back