How Notion’s design team uses Claude Code to prototype | Brian Lovin (Product designer)
Brian Lovin, a product designer at Notion, demonstrates how Notion's design team uses Claude Code to rapidly prototype AI-powered features by building a shared "Prototype Playground"—a Next.js repository where designers can quickly iterate from Figma designs to working code. The episode showcases how AI-assisted coding tools dramatically reduce friction in design-to-implementation workflows, particularly for AI products where static mockups fail to capture the actual user experience.
Key takeaways
- • Prototype Playground is a shared Next.js app with per-designer namespaced directories and shared components, eliminating the need for individual repos and enabling designers to discover and reuse each other's ideas.
- • When Claude asks you to do something (e.g., "check the browser"), teach it to perform that task itself using MCP tools and automation scripts, enabling longer, more autonomous runs without human intervention.
- • Use Claude's Plan Mode before execution to catch logical errors upfront, read the plan carefully, and provide good context via root-level `.claude.md` files and local `.claude.local.md` configurations for team collaboration.
- • AI-assisted Figma-to-code conversion using the Figma MCP can achieve ~80% accuracy with automated verification loops that compare browser output to design files, reducing manual handoff work significantly.
- • Create Claude Skills and slash commands that bundle natural language instructions with executable scripts to handle repetitive problems like icon name hallucination, Figma imports, and multi-step deployment workflows.
- • Code-first prototyping is essential for AI products because static designs cannot capture real model behavior, loading states, error handling, and user confidence during multi-step interactions that require live experimentation.
- • Design workflows benefit from Opus 4.5's superior planning, execution, and communication style, which combined with structured tools (terminal UI, side-by-side windows) and good sleep/rest produces dramatically better prompting outcomes.
Recommendations (11)
"All of a sudden these AI coding tools come along and now I can prototype faster. I can prototype in production."
Brian Lovin · ▶ 4:11
"You can give it access to MCP tools. So, Playwright is one, the Chrome Dev Tools MCP is another one."
Brian Lovin · ▶ 19:28
"You can give it access to MCP tools. So, Playwright is one, the Chrome Dev Tools MCP is another one."
Brian Lovin · ▶ 19:28
"I just love the GitHub Desktop app. It just like it gives you buttons for all of this. You can see your diffs."
Claire Velo · ▶ 37:44
"We have a cloud.md file at the root of our project with just some rough instructions around like the tooling that we use like we use bun, we use Tailwind."
Brian Lovin · ▶ 16:29
Mentioned (2)
More from these creators
Claude Code for personal productivity: a step-by-step tutorial | Hilary Gridley
How Stripe deploys 1,300 AI-written PRs per week
How Microsoft's AI VP automates everything with Warp | Marco Casalaina
From journalist to app developer using Claude Code
How Figma engineers sync designs with Claude Code and Codex
5 steps to generate consistent brand images with Midjourney