• AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
• multi-agent — Using several AI agents on one problem — splitting work in parallel, checking each other, or filling different roles like planner and reviewer.

Lays out concrete techniques for agents that reason about two-dimensional space, an area where text-trained coding agents typically fail, plus a released MIT starter kit to build on.

• 0:00A demo kicked off before the talk begins
• 1:08tldraw, the app and the SDK behind other canvases
• 4:14Why coding agents work and canvas agents struggle
• 5:08Teaching a model to read and act on the canvas
• 7:01The agent starter kit goes agentic
• 8:26Fairies, multiplayer agents you can see working
• 9:49A group chat of agents with an orchestrator
• 12:06One node per coding agent, in a dependency graph
• 15:48Code mode, a window manager, and Pong
• 18:05The canvas as a place to work with agents

• videoMousepower: agents that can’t be measured, can’t be managed. — Maximillian Piras, Yutori
• videoOne Designer + AI. Hundreds of Deliverables. — Vincent Wendy, AI Engineer
• videoIt’s Tokens All The Way Down: How RLMs are Different — Kevin Madura, AlixPartners
• videoYour agents lack context: Here's how to fix "You're absolutely right!" — Brandon Waselnuk, Unblocked

• blogHow canvases make agentic workflows visible, steerable, and cost-efficientAyan Gupta
• videoThe Dark Arts of Web Automation: Teaching Agents to Use Websites Like Humans — Corey Gallon, RexmoreAI Engineer
• videoAgent Frameworks Considered Harmful — Rémi Louf, .txtAI Engineer

Checking sign-in…

Loading comments…