Max Drake

Drake studied mechanical engineering at MIT, earning a bachelor’s degree, and previously worked at Fermat across product research, design and engineering. Work during the MIT period included the Tangible Media Group, the Self-Assembly Lab, propelland and Mercedes Benz R&D. At tldraw, Drake combines product engineering with experiments in how people and agents can share a visual workspace.[1][5]

Give agents a place to look and act

Drake’s spatial-harness article traces a series of tldraw experiments. The system supplies screenshots and structured shape data, provides canvas-editing operations and lets agents move their own viewport to inspect more of the document. Later iterations retrieve context when needed and allow code to manipulate the editor. A canvas linting system can warn about problems such as overlapping text shapes.[2]

Record an animation by performing it

Drake coauthored and helped build tldraw flash with Audrey Leo and Leo Knudsen. Users draw a scene and record movements by dragging objects. The tool records other canvas changes as well, including drawing and text edits. Existing shapes can be repositioned after recording, and customizable characters retain their underlying motion. The project builds on tldraw’s SDK mechanics to make short animations directly on the canvas.[3]

In the conference talk, Drake demonstrates fairies, recognizable agent characters that share a canvas and delegate work. Their appearance distinguishes one agent from another, while animation communicates whether they are active or waiting. A later prototype places coding-agent tasks in a dependency graph, connecting an overview of work with actions outside the drawing environment. The examples explore interface behavior and coordination rather than establish that every generated result is correct.[4]