tldraw computer

Visual workflows on an infinite canvas, run by a language model

Intermediate
Screenshot of tldraw computer, Visual workflows on an infinite canvas, run by a language model

What is tldraw computer?

tldraw computer is an AI project builder tool. An experimental canvas from the tldraw team where connected components form workflows, and a multi-modal language model executes them as a runtime. Branching and looping are supported, and six example projects can be remixed into your own account.

tldraw computer is an experimental project from the team behind the tldraw SDK and the tldraw.com collaborative whiteboard. It turns the infinite canvas into a programming surface: users place components, wire them together, and let a multi-modal language model act as the runtime that executes each instruction. The pitch on the site is natural language computing, where a workflow is drawn rather than typed in code. The build loop is short. A component is created and connected to its neighbours, then run on its own to produce data. Outputs flow along the connections, so a downstream component receives whatever upstream produced and transforms it further. Because the model handles interpretation, the components can hold plain language instructions instead of scripts. Workflows can also branch and loop, which moves the tool past simple prompt chaining into something closer to a small program with conditional paths. Six pre-built examples show the range: a Story Generator, a Sorting Machine, a Battle Simulator, an Invention Inventor, an Email Factory and an Animal Lab. Each opens as a starting project that can be edited into a new copy stored in the user's account. That makes the examples templates as much as demos, and it is the fastest route for anyone who wants to learn the component model by pulling an existing graph apart. This is a tool for tinkerers, designers and technically curious people rather than a production pipeline. The site presents it as an experiment, so expect rough edges and a learning curve around how components should be phrased and sequenced. Alternatives split in two directions: node-based workflow builders, which demand real logic but give deterministic control, and plain chat interfaces, which are easy but linear. tldraw computer sits between them, offering spatial structure without formal syntax. It runs free in the browser.

How do you use tldraw computer?

  1. 1Open an example project
    Pick one of the six pre-built workflows on the homepage, such as the Story Generator or the Sorting Machine. Each one loads as a working graph you can inspect before changing anything.
    tldraw computer — Open an example project
  2. 2Inspect components and their wiring
    Trace the connections between components to see how data moves through the graph. This is the quickest way to understand what each component is instructed to do.
  3. 3Run a single component
    Execute one component in isolation to generate its data and check the output. Confirming each stage before expanding the workflow keeps debugging manageable.
  4. 4Build a branching workflow
    Add your own components, connect them, and set up paths that branch or loop. Because the language model acts as the runtime, instructions are written in plain language rather than code.
  5. 5Save an edited copy
    Editing an example creates a new copy in your account. The original template stays intact, so you can return to it or start a second variant.

Pros and cons

Pros

  • Workflows are drawn spatially, so branching and looping logic stays readable at a glanceAI
  • Six ready-made example projects (Story Generator, Email Factory, Animal Lab and others) act as editable templatesAI
  • A single component can be run on its own to inspect the data it produces before wiring more of the graphAI
  • Plain language instructions replace code, which lowers the barrier for non-programmersAI
  • Free to use in the browser with nothing to installAI

Cons

  • The site labels the project experimental, so stability and long term support are not guaranteedAI
  • No pricing, integration or API documentation is published, which makes it hard to judge fit for serious workAI
  • Outputs depend on a language model, so results vary between runs and are not deterministicAI
  • Workflows live in a tldraw account, which ties projects to one ecosystemAI
  • Beyond the example templates there is little onboarding material for designing a graph from scratchAI

How much does tldraw computer cost?

Free trial

No separate trial is needed because the tool is described as free to use in the browser.

Learn more

Pricing

The site states the tool is free in your browser. No paid tiers, usage limits or subscription details are published on the homepage.

Learn more

Features

Components are placed and connected on an infinite canvas, and a multi-modal language model runs them as a runtime. Each component can be executed alone to generate data, outputs pass between connected parts, and workflows support branching and looping. Six example projects (Story Generator, Sorting Machine, Battle Simulator, Invention Inventor, Email Factory, Animal Lab) can be opened and edited into a personal copy.

Frequently asked questions about tldraw computer

  • How much does tldraw computer cost?
    The site states the tool is free in your browser. No paid tiers, usage limits or subscription details are published on the homepage.
  • Does tldraw computer offer a free trial?
    No separate trial is needed because the tool is described as free to use in the browser.
  • How do you use tldraw computer?
    The walkthrough on this page covers 5 steps: 1. Open an example project 2. Inspect components and their wiring 3. Run a single component 4. Build a branching workflow 5. Save an edited copy.
  • What platforms does tldraw computer support?
    tldraw computer is available on Web App.
  • What are the limitations of tldraw computer?
    The site labels the project experimental, so stability and long term support are not guaranteed. No pricing, integration or API documentation is published, which makes it hard to judge fit for serious work. Outputs depend on a language model, so results vary between runs and are not deterministic.

Status

StatusActive
Views0
Outbound clicks0
Added10/21/2026

Platforms

Web App

Pricing

Free