WhoBuilt

How To Stand Out In A World Of AI Designs

Oct 22, 2026· 48 min· 20K views·Watch on YouTube

Speakers

Alex Danalovich, co-founder and CEO of Magic Patterns, joins Y Combinator’s Design Review to demo the AI design tool and discuss product design, customer feedback, and how to make startup websites clearer.

From front-end engineering to Magic Patterns

Danalovich and his co-founder were front-end engineers who often had to turn Figma mockups into code. After YC advised them to build around what they knew, they spent about nine months exploring ideas before Magic Patterns found users who paid and returned.

A shared canvas for design and code

Danalovich demonstrates Magic Patterns’ canvas, where teams can explore design variations and work with images, documents, and prompts. Its MCP connection lets people use context from tools such as Cursor or Claude Code to create and share prototypes. He says the canvas is not right for every user, and onboarding needs to help people decide when to use it.

Make landing pages clear and easy to use

Reviewing startup sites, Danalovich and the host point out unclear calls to action, hard-to-understand product descriptions, misleading interactive elements, and scroll-jacking. They recommend stating what a product does and who it serves near the top, making the next action obvious, and using animation to explain the product. They also highlight clear privacy information for AI products that handle sensitive data.

Build taste through users and craft

Danalovich recommends studying design, including reading The Design of Everyday Things, while keeping the user experience at the center. Everyone at Magic Patterns handles customer support, so the team sees problems firsthand and fixes them. He encourages teams to polish rough AI-generated work, arguing that clearer design can help users understand and complete their tasks.

Magic OS and code as a medium

Danalovich describes Magic OS as a canvas for combining design work with product context such as PRDs. He sees code as an enduring medium for production, while acknowledging that the direction of design tools is changing too quickly to predict with certainty.

Takeaways

  • Tell visitors what your product does, who it helps, and what to do next near the top of the page.
  • Keep calls to action visible and straightforward; avoid interactions that look usable but do not work.
  • For AI products handling sensitive data, explain what data is accessed, how it is used, and how it is protected.
  • Bring your design system and company context into AI tools before prompting to reduce repeated corrections.
  • Have the whole team hear customer problems directly, then make time to improve rough first drafts.

Mentioned in this talk

Search inside this talk

What was claimed 17

Related talks