Steve Ruiz

Steve Ruiz is the founder and chief executive of tldraw, the London company building an infinite-canvas SDK for interactive web applications. A visual artist turned product designer and developer, he has transformed the digital whiteboard into infrastructure for sketching software, running visual workflows and collaborating with visible AI agents.

Ruiz earned an MFA in visual art from the University of Chicago in 2013 and began his career as a painter before moving into technical product design. He worked at ClearScore, created educational material at Framer and became a product designer at Play.

His early open-source tools addressed the visual mechanics that make digital drawing feel natural. perfect-freehand generates pressure-sensitive strokes from pointer movements; perfect-arrows creates visually coherent arrows between points and shapes. Implementing digital-ink functionality in Excalidraw convinced him that developers needed reusable foundations for spatial interfaces, including drawing primitives, interaction states and collaborative editing.

Ruiz started tldraw as an open-source project in 2021 and founded the company in 2022. Its React-based SDK powers custom canvas applications, while the free tldraw whiteboard demonstrates the technology. In April 2025, tldraw raised a $10 million Series A after generating $1 million in first-year licensing revenue from customers including ClickUp and Padlet. SDK 4.0, released that September, added starter kits for agents, visual workflows, branching conversations and multiplayer applications.

A canvas where software and agents become visible

• Make Real: Introduced in 2023, Make Real converts sketches and annotations into functioning web interfaces. Generated applications stay on the canvas, so users can mark defects, revise controls and feed those visual changes into subsequent iterations. Ruiz has used the approach to create a stop-motion application with camera input and onion-skinning.

• tldraw computer: Built in collaboration with Google, tldraw computer connects prompts, images, camera feeds and speech inside executable visual graphs. Multimodal models interpret mixed inputs, while branching, loops and persistent results support creative experiments and potential approval-based business workflows. Google provides model capabilities; tldraw supplies the interface and workflow environment.

• Visible multi-agent collaboration: Fairies places AI agents directly on a shared canvas, making their locations, actions and overlapping work observable. Agents can elect a leader, delegate assignments and edit compositions concurrently. Their outputs remain editable shapes and text, exposing practical coordination problems involving shared state, conflicting edits and incompatible spatial conventions.

• User-controlled agent execution: Ruiz has experimented with giving locally running agents programmatic access to a desktop canvas, allowing diagrams and interactive software to modify each other. His on-canvas agent work treats unrestricted code execution as a genuine security risk: deeper agent capabilities require explicit decisions about local files, browser access and user control.