Computaco

ComputacoA world of computers.

Machines, apps, networks and the people who use them, in one simulation you can run, fork and measure.

Every machine above is running in your browser.

35 µsto fork a whole world
4.8 µsto run a terminal command
5operating systems on one network
260verified expert episodes
14professional domains

However you want.

Choose the machines, their operating systems and apps, the network between them and the services they reach. Start from a built-in world or define your own.

Create a world
  • Ubuntu, a terminal

  • macOS, FreeCAD

  • Windows, Excel

  • Ubuntu, KiCad

  • macOS, iMovie

  • Windows, Slack in Edge

  • Ubuntu, GIMP

  • macOS, GitHub in Safari

  • Windows, Paint

  • Ubuntu, Firefox

  • Release engineers, 3 desktops: each machine with its agent’s conversation beside it, at the end of the run

    Release engineers, 3 desktops

  • Hardware engineers, 3 desktops: each machine with its agent’s conversation beside it, at the end of the run

    Hardware engineers, 3 desktops

  • Data and finance, 3 desktops: each machine with its agent’s conversation beside it, at the end of the run

    Data and finance, 3 desktops

  • On call, 3 desktops: each machine with its agent’s conversation beside it, at the end of the run

    On call, 3 desktops

  • Planning the launch, 3 desktops: each machine with its agent’s conversation beside it, at the end of the run

    Planning the launch, 3 desktops

  • Design and docs, 3 desktops: each machine with its agent’s conversation beside it, at the end of the run

    Design and docs, 3 desktops

  • Northstar on launch day, 5 desktops: each machine with its agent’s conversation beside it, at the end of the run

    Northstar on launch day, 5 desktops

  • Northstar at month end, 3 desktops: each machine with its agent’s conversation beside it, at the end of the run

    Northstar at month end, 3 desktops

  • Halyard Robotics engineering, 4 desktops: each machine with its agent’s conversation beside it, at the end of the run

    Halyard Robotics engineering, 4 desktops

  • Halyard Robotics sales and launch, 3 desktops: each machine with its agent’s conversation beside it, at the end of the run

    Halyard Robotics sales and launch, 3 desktops

  • Halyard Robotics purchasing, 3 desktops: each machine with its agent’s conversation beside it, at the end of the run

    Halyard Robotics purchasing, 3 desktops

  • Lantern Media newsroom, 3 desktops: each machine with its agent’s conversation beside it, at the end of the run

    Lantern Media newsroom, 3 desktops

  • Lantern Media production, 2 desktops: each machine with its agent’s conversation beside it, at the end of the run

    Lantern Media production, 2 desktops

  • 6 people, 6 phones, 1 rumor: each machine with its agent’s conversation beside it, at the end of the run

    6 people, 6 phones, 1 rumor

  • 8 people, 8 phones, 1 rumor: each machine with its agent’s conversation beside it, at the end of the run

    8 people, 8 phones, 1 rumor

  • 6 more people, 6 phones, 1 rumor: each machine with its agent’s conversation beside it, at the end of the run

    6 more people, 6 phones, 1 rumor

  • 6 buyers, 6 phones, 1 story: each machine with its agent’s conversation beside it, at the end of the run

    6 buyers, 6 phones, 1 story

Put people in it.

Each person has a phone or a computer, sees only their own screen and acts on it. A script, a language model or your own agent decides what they do. One built-in society is a company and its buyers on launch day: 21 people.

Read about societies

6 people, 6 phones, 1 rumor

Simulate everything.

Real work crosses people, machines and days. Each job here was done through the screen, with pointer and keyboard, and judged by checks on the world’s state.

See the expert work

Rehearse a change.

Fork the world at a checkpoint and change a rule, a setting or an action. Every branch starts from the same state, and the same actions always reach the same state. A fork takes 35 µs.

The checker forks at every action an agent may take, to the depth you set, and returns the shortest sequence that breaks a policy. On the reference world it checked 77,748 states and broke 9 of 15 policies.

See the benchmark
Read the policy checks
The console's Analytics pane The policy check: 6 of 15 policies held and 9 were broken over 77,748 states (9,473 distinct) in 179 minutes, with 9 counterexamples replayed and verified; one row per policy, P1 to P15, each marked held or violated with the length of its counterexample.

Verified expert computer-use trajectories for 3,811× less than the cheapest computer-use agent.

We take a job a professional already does, such as closing the month’s books or drawing a part in CAD, and do it on a simulated computer with pointer and keyboard. The job’s own checks confirm the result, so every recording is a correct example to train on. The same jobs done through direct tool calls take fewer steps: across 260 recordings, the screen took 4.8× as many actions.

Every domain and episode
Tool callsGUI actions

The chart draws from data/expert.json.

Turn paid-per-seat slow software into fast free simulations.

We record an application at work, label every frame with its text, its controls and their bounds, rebuild it as a deterministic simulation, and align the two step by step.

94.6%KiCad, 2,586 steps
87.1%VS Code, 296 steps
81.4%FreeCAD, 3,144 steps

Must be lawfully acquired, Sega v. Accolade (1992) and Google v. Oracle (2021)

A rendered frame: Safari on macOS showing a GitHub pull request. The same frame with a box around each of its 96 text runs and 67 enabled controls.

Simulate the market.

Put a product, a price or a pitch in front of simulated buyers, readers and executives, and measure what they choose. Each one has a persona, a budget and a screen of its own.

All research

Computatrum.

An agent that uses your own computer while you watch. It does the work that has no API, in any app you already have.

Download Computatrum

ComputaCo API

Build simulated computers, offices and customers into your own product. Your team or your agents create worlds, run work in them and collect verified results, on our hosted machines or your own, from Python, JavaScript or Rust.

pip install computerworld==0.3.0 --find-links https://github.com/ComputaCo/computerworld/releases/expanded_assets/engine-v0.3.0
from computerworld import World

world = World(definition, seed=42)
env = world.environment({"actor": "alice", "machines": ["alice-mac"],
                         "actions": ["pointer.v1", "keyboard.v1", "application.v1"],
                         "observations": ["semantic.v1"]})

env.step([{"family": "application.v1", "op": "launch",
           "machine": "alice-mac", "payload": {"kind": "code"}}])

scene = env.scene(1440, 900)     # roles, names, geometry — no pixels needed
frame = env.render(1440, 900)    # or exact RGBA, if your model wants pixels