Dive Club 🤿 – Steve Ruiz: Is the Canvas the Future for AI?

Host: Ridd
Guest: Steve Ruiz (Founder, TLDraw)
Release Date: December 1, 2025

Episode Overview

This episode features an in-depth conversation with Steve Ruiz, founder of TLDraw, a pioneering canvas SDK powering numerous startups at the intersection of design, real-time collaboration, and AI tooling. Host Ridd and Steve explore the evolution of the digital canvas, the nuances of tool design, the nature of collaborative design environments, and what the future holds for collaborative AI on canvases.

The discussion also delves deeply into Steve’s personal journey from fine arts to developer to founder, the creative and technical obsession with seemingly small problems (like drawing the perfect arrow), and practical theory on how the next generation of AI-first tools will operate visually.

Steve Ruiz’s Journey to Toolmaking

From Fine Art to Tech (01:15–08:26)

• Educational Roots: Steve starts with studying fine art at university and grad school, emphasizing not just craft but the academic/industry context of visual creation.
• Transition to Design: Facing the fragility of creative careers, moving from the US to the UK forced Steve to rethink his professional path. He gravitated towards publishing design and quickly became immersed in digital production (Adobe InDesign, then ebooks and web design).
• Early Prototyping: Inspired by web design’s evolution around 2016, he leaped toward prototyping with early tools like Framer Classic and Origami, learning coding in CoffeeScript to create interactive design prototypes.

“If you got in early with those tools, it really was a differentiating factor... you had to have an appetite for complexity back then.”
— Rid (06:00)

Obsessing over Micro-Problems: Arrows & Drawing (10:19–17:43)

• Steve shares how a random pizza-slicing math problem from his youth evolved into a lifelong fascination with visual computation and geometric problems, translating to obsessing over “perfect arrows” in design tools.
• His openness about failure—struggling to generate interesting teaching content, iterating on the perfect vector implementation—highlights the trial-by-fire learning process of building tools that underpin the creative process.

“[Drawing arrows] became the most complicated thing that I'd ever tried to do with code... It was turned out to be highly subjective.”
— Steve Ruiz (13:26)

Key Principles of Good Tool Design

Tools as Decision-Making Engines (24:58–26:41)

• Decision Primacy: Steve views all design tools (e.g., color pickers) as fundamentally about enabling and accelerating decision-making.
• Safety & Comparison: Great tools must offer safety nets (undo, non-destructive editing) and quick comparison of alternatives.
• Balance of Precision: The best tools balance letting designers be precise when necessary but not at the cost of speed or cognitive load.

“A good tool will allow you to... very safely make a change and go back, have that safety net... You also need to just be able to compare options.”
— Steve Ruiz (25:11)

The Challenge & Promise of the Canvas

Conventional Wisdom vs. Innovation (27:23–30:30)

• Canvas as Commodity: Many conventions (selection, zoom behavior, undo) are now expected “commodities” in digital canvases—get these wrong, and the product feels broken.
• Balancing Familiarity and Progress: The tension is knowing which behaviors/tools must remain conventional and where true innovation can happen.

“If it doesn't work like that, it feels broken. It feels like, oh, this is like a tech demo. This is not a real thing.”
— Steve Ruiz (28:31)

Why the Canvas? (31:51–36:22)

• Expanding Use Cases: Modern canvases aren’t just digital whiteboards. They're vehicles for workflows, education, team onboarding, even verticals like pharmaceuticals.
• Multiplayer & Verticalization: The multiplayer canvas experience (real-time, multi-human, and potentially multi-AI collaboration) is still under-explored, and the canvas is evolving toward more vertical, specialized use-cases.

“So many entire kind of product categories can fit into the canvas, but the ones that we're familiar with are... an early generation.”
— Steve Ruiz (34:54)

Canvas + AI: The Experiments So Far

Make Real, Generative Demos, & Multi-Agent Canvases (37:56–46:06)

• Steve details the virality and appeal of TLDraw + generative AI demos—draw a website, click ‘Make Real’, and get a working prototype, all powered by image-understanding models.
• Multi-Agent Collaboration: Steve discusses experiments with AIs as “collaborators” on the canvas—some as chatbots, some as virtual “fairies” (an intentional metaphor) with their own tasks, identities, and behaviors.

“For a lot of people, this was their first time making something that works, like producing an artifact of software in any kind, any shape. But this madness... there were like millions and millions of billions of views.”
— Steve Ruiz (40:25)

• Fairies Metaphor: Experimenting with how to give autonomous agents distinct identities on the canvas (visual design, posture, “gifts,” context) makes the orchestration of complex workflows with multiple AIs much more tangible and comprehensible.

“They’re not people, they’re definitely not people, but they’re... kind of humanoid. They're not, like, bonded to you... They have these powers and they have accessories...”
— Steve Ruiz (46:58)

The Cutting Edge: Multi-Agent Orchestration with Canvas

What’s Next – AI Collaboration UI (43:21–61:40)

• Collaboration Vision: The team’s guiding thesis—collaborating visually on a canvas is already excellent for human teams, so it should work for AI too.
• Technical Hurdles: While current models struggle with precision and context (“bad” bots, coordination challenges), the visual medium still excels at managing, observing, and steering multiple agents (bots and humans) simultaneously.
• Metaphors Fuel Innovation: Both technical (state machines akin to video game AI) and metaphorical concepts (fairies, enchanted ponds) drive rapid idea generation and system design.

"One of the biggest problems... is that it's very hard to remember which agent is doing what... that actually works really well on the canvas because they just look different, you know, give them different hats, give them different wing patterns and whatever.”
— Steve Ruiz (54:28)

• Inspiration from Video Games: Multi-agent orchestration draws from video game AI strategy (agents with states moving between “waiting,” “acting,” “patrolling”), adapted to collaborative productivity contexts.

Notable Quotes & Memorable Moments

• On Tool Design:
  “If you have no choice but to make every decision a precision decision, then... the speed of operating just becomes really slow.”
  — Steve Ruiz (26:12)
• On Canvas Convention:
  “If you get one of those really important conventionalized features wrong... it's just completely unusable.”
  — Steve Ruiz (28:14)
• On Emergent Use-Cases:
  “The multiplayer canvas experience is so good and so under explored...”
  — Steve Ruiz (34:26)
• On Multi-Agent Collaboration:
  “How do you handle orchestration, how do you handle waiting, turn taking, or following the different states that the AI is going to be in?”
  — Steve Ruiz (59:36)
• On the Future of Collaboration:
  “We are in massively uncharted territory but also like some of the most ambitious stuff happening in software...”
  — Steve Ruiz (59:59)

Important Timestamps

• Overview of Steve’s Background: 01:15–08:26
• The Pizza Problem & Obsessing Over Arrows: 10:19–13:26
• Principles of Good Tool Design: 24:58–26:41
• Commodity Nature & Convention in Canvases: 27:23–30:30
• The Canvas in Modern Tooling: 31:51–36:22
• Generative AI + TLDraw Demos: 37:56–42:52
• Birth of “Fairies” & Multi-Agent Canvas Orchestration: 46:58–56:11
• Metaphor in Tool Design & Orchestration: 57:24–59:24
• Comparisons to Video Game AI: 59:24–61:40

Key Takeaways for Designers & Builders

• Great tools empower safe, rapid decision-making by balancing precision & flexibility.
• Infinite canvases will play a pivotal role in future AI-powered collaboration, enabling real-time, multi-agent, and human-AI synergy in previously impossible workflows.
• Getting foundational UX conventions right is non-negotiable; users expect the “commodities” of interaction, or the tool instantly feels untrustworthy.
• AI “collaborators” benefit from visual identity, context, and metaphor—turning bots into fairies is more than whimsy: it provides concrete UX affordances for managing complexity.
• The best way to shape the future of tools is to experiment publicly, share learnings, and embrace a blend of foundational theory and creative metaphor.

For Further Exploration

• Follow Steve Ruiz and TLDraw on Twitter/X for real-time demos and development updates.
• Check out the layered demos at teachdra.com.
• Explore the Dive Club archive at dive.club for more content on the intersection of design, AI, and tooling.

This episode is essential listening for anyone interested in tool-making, generative AI, collaborative environments, and the rapidly evolving role of the canvas in product and workflow design.