Blog is back tldraw.dev/blog/back-to...
14.01.2026 20:19
👍 12
🔁 1
💬 0
📌 0
You too can impress your customers, friends, and future friends by building canvas experiences with the tldraw SDK. Visit tldraw.dev to learn more.
Welcome @Geniolearn as the newest customer for the tldraw SDK! Genio creates learning tools for students and uses tldraw for visual note-taking in their Genio Notes app on web and mobile.
You too can build canvas features and become a customer at tldraw.dev.
Welcome @snaptrude as the latest customer for the tldraw SDK. Snaptrude uses tldraw to power its present mode where users share and collaborate around architectural designs.
And if you want to build something like this, we have a starter kit for building agents that control the canvas. Check it out here: tldraw.dev/starter-kit...
Check out tldraw.com to try it yourself. We're unshipping the fairies at the end of the month, so get em while you can!
We also encouraged the agents more to request new context. This would create a new prompt to continue what they were doing before, but with new context about the canvas, todo list, and fairies, too.
You can see this if you select three fairies and prompt the group. One fairy will be elected as the orchestrator, create the project, and wait for the other fairies to finish the work.
In earlier builds, fairies had different personalities and talents. The orchestrator fairy would assign tasks based on the most appropriate worker fairy available, preferring to use a more creative fairy for some tasks and more operationally inclined fairies for other tasks.
We ended up using a 'fairy management system', where one fairy would draft the plan, create the todos, assign the todos to other agents, and then coordinate their output with followup todos.
For example, if three fairies began working on a wireframe with todos for a header, body and footer, they would often place these in completely arbitrary places relative to the others.
There were many issues with this. What if two agents start working on the same task? Even if agents work on different tasks, how do they work together towards an overall goal? And in doing this, how do they coordinate between different parts of the overall goal?
Our first attempt at orchestrating multiple fairies was to create a shared todo list for all of the agents and give it as context alongside screenshots and structured data about the canvas.
How do you coordinate when agent spend most of their time 'underwater' and only briefly and intermittently ‘return to the surface’ for context?
In a multi-agent system, things can change a lot while an agent is working. Another agent might be editing the canvas. Humans might be moving things around.
For example, on tldraw, one person can draw something while speaking to their collaborator and looking at what they are drawing. An agent can only see the state of the canvas or communicate with other agents before and after it has finished drawing.
Agents only receive new context when they're prompted. While they stream back their response, they can't get new context. So unlike humans who actively observe, communicate, and learn while working collaboratively, agents are essentially working blind. No real-time feedback.
Getting multiple AI agents to work together on an infinite canvas means tackling a huge coordination problem. Here's what we learned while building our fairies feature (december only on tldraw dot com)
The @inkandswitch.com team organised a social at the @tldraw.com office in London to share some recent news and research from the lab.
priti.is/chasingspark... 🌱
My favourite moment: bonus Tldraw experiments from @todepond.com
Huge
Grab a starter kit, build something incredible, and let us know what you've made! Follow here, star us on GitHub, or join the Discord, links in bio.
Remember, tldraw is a company and you can work with us. Drop us a message at tldraw.dev/careers. Here's our office.
We've also updated our licensing model for the tldraw SDK. From 4.0 forward, the SDK will require a license key to work in production. You can grab a free 100-day trial license at tldraw.dev/pricing. Non-commercial projects may apply for a hobby license for unrestricted use.
In addition to starter kits, this release completes the months-long effort to make the tldraw SDK compliant with WCAG 2.2 AA. Paperwork coming soon.
In practice, this means better keyboard controls, a clearer, more customizable UI, and accessibility improvements for all of our users and customers.
5. The branching chat starter kit features a threaded AI chat where the user can branch off from the conversation in multiple directions.
We built this on top of our workflow starter kit in an afternoon!
4. The chat starter kit puts an inline canvas into a chat application so users can draw, add markup and include additional context to regular chat messages. Every AI chatbot should have this. Show your boss. Tag @cursor_ai and tell them to put the canvas in the chat ffs. Ask for tldraw by name.
