Cursor Head of Design Reviews Startup Websites
Timestamps are as accurate as they can be but may be slightly off. We encourage you to listen to the full context.
Podcast Summary
Cursor Head of Design Ryo Lu joins Design Review to analyze user-submitted websites built with Cursor AI, offering unfiltered feedback on design choices, messaging clarity, and user experience. (00:00) The episode walks through eight different startup websites, examining everything from hero sections and value propositions to visual design and conversion optimization. Ryo doesn't hold back, calling out common AI-generated design pitfalls like purple gradients, inconsistent styling, and jargon-heavy messaging while providing actionable advice for creating more effective product websites.
- Core theme: How to communicate what your company does clearly and effectively through website design, avoiding common AI-generated design traps
Speakers
Ryo Lu
Head of Design at Cursor, the leading AI coding tool used by more than a million people worldwide. Previously served as a founding designer at Notion and worked as a product designer at Stripe and Asana. Known for his expertise at the intersection of design and engineering, having built fan sites as a kid before moving into professional product design roles.
Aaron Epstein
Host of Design Review from Y Combinator, focusing on helping startups improve their product communication and website design through expert analysis and feedback.
Key Takeaways
Make Your Value Proposition Crystal Clear in 5 Seconds
Throughout the episode, Ryo consistently emphasized that visitors ask themselves three key questions when they hit a website: "What is this? Is it for me? Does it work?" (08:07) Many of the reviewed sites failed this basic test by using vague headlines and industry jargon instead of clear, direct explanations. For example, Clavis AI used terms like "progressive discovery" and "smart navigation" that meant nothing to users, while Velvet only had six words describing what they actually do. The most effective approach is to use language your target customers already understand, speaking to their existing problems rather than introducing new concepts they need to learn.
Avoid the "AI Slop" Visual Design Trap
Ryo identified common visual patterns that immediately signal AI-generated design: massive shadows, purple gradients, purple buttons, and poor typography choices. (05:00) These default AI styling choices make websites look generic and unprofessional. Instead, he recommends going with safer, cleaner options like system fonts and well-established design foundations. The key is building robust foundational design tokens and components that AI can then compose effectively, rather than letting AI make all the styling decisions from scratch.
Prioritize Single Call-to-Actions Per Section
Multiple sites suffered from "CTA chaos" - too many buttons, links, and competing actions that confused visitors about what to do next. (09:19) Clavis AI had join Discord, check GitHub, YC backing, product hunt badges, start free buttons, documentation links, and sign-up buttons all competing for attention. Ryo's rule is simple: every scroll should have only one main CTA with clear priority hierarchy. This focuses user attention and improves conversion by eliminating decision paralysis.
Show, Don't Just Tell Your Product Value
The most successful sites in the review were those that let users experience the product immediately rather than just describing it. Freya stood out because visitors could actually talk to their AI voice assistant right on the homepage. (22:58) Similarly, Ryo suggested that instead of listing integrations in text, companies should visually show all the tools they connect to, making the value immediately obvious. This "show don't tell" approach is especially important for technical products where the capability might be hard to understand from description alone.
Maintain Design Consistency Across Your Entire Site
Several sites had jarring inconsistencies between their marketing pages and actual product interfaces, with completely different color schemes, typography, and visual styles. (16:18) CodeCrafters exemplified this problem - their marketing site used purple gradients while their product interface was black and white, creating a disjointed user experience. Maintaining visual continuity builds trust and creates a more professional impression, signaling that the company pays attention to important details.
Statistics & Facts
- Cursor is used by more than a million people worldwide, establishing its credibility as a leading AI coding tool. (00:05)
- Freya's AI voice assistant can handle conversations in over 40 languages and operates 24/7, demonstrating the scale of their voice AI capabilities. (23:15)
- One customer mentioned in Finta's testimonials was able to claim $17,000 in tax credits through their automated process, showing concrete financial value. (28:22)
