Y Combinator's president reviews real AI startup websites, revealing critical design mistakes and best practices for communicating breakthrough technology to potential customers.

Key Takeaways

• Don't be clever if clever equals confusing - clear value propositions beat witty taglines when explaining new AI categories
• Focus on one primary call-to-action rather than overwhelming visitors with multiple product offerings and competing buttons
• AI startups must demonstrate "feats of strength" through videos or demos showing capabilities users have never seen before
• Be specific about your ideal customer profile rather than targeting "anyone who might need our technology"
• Use animation sparingly and purposefully to direct attention, not create sensory overload with moving elements everywhere
• Social proof from recognizable companies and detailed case studies build essential trust for enterprise AI products
• Lead with concrete benefits and savings rather than technical features when positioning AI tools
• Spacing and visual hierarchy mistakes undermine trust in enterprise-level AI solutions

Timeline Overview

• 00:00–01:01 — Intro: Introduction to the AI startup website design review with YC President Garry Tan, focusing on the challenge of describing new AI categories
• 01:01–05:51 — Rosebud: Critique of AI game asset generation platform struggling with multiple products, confusing messaging, and competing call-to-actions
• 05:51–10:41 — Magicflow: Review of AI workflow platform that promises drag-and-drop functionality but creates trust issues with misleading interactions
• 10:41–15:44 — Reality Defender: Analysis of deepfake detection company with unclear value proposition and poor visual hierarchy spacing issues
• 15:44–20:09 — Pump: Examination of AWS cost reduction service using excessive animation and failing to differentiate from competitors
• 20:09–23:35 — Voiceflow: Positive review of voice/chat assistant builder demonstrating effective design principles and clear value communication
• 23:35–End — Outro: Summary thoughts and invitation for founders to submit websites for future review episodes

The Challenge of Describing New AI Categories

AI startups face a unique challenge that traditional software companies don't encounter: explaining capabilities that literally didn't exist before. This fundamental difference requires different approaches to website messaging and demonstration.

• The "why now" moment for AI startups centers on breakthrough capabilities that large language models have recently enabled. This creates both opportunity and confusion as founders must educate markets about possibilities that weren't available months or years ago.
• Category creation becomes essential when building AI products because existing mental models don't capture what these tools can accomplish. Founders must simultaneously introduce new concepts while making them immediately understandable to potential customers.
• Technical jargon multiplication happens naturally as founders immerse themselves in AI terminology, but this creates barriers for users who need simple explanations of complex capabilities. The more sophisticated the technology, the simpler the explanation needs to be.
• Demonstration over description becomes critical for AI products because people need to see capabilities in action rather than trying to imagine them from text descriptions. Static explanations rarely capture the magic of AI-powered features.
• Reference points matter when introducing new categories - connecting AI capabilities to familiar concepts helps users understand value without requiring deep technical knowledge about machine learning or natural language processing.

The key insight is that AI startups must balance showcasing revolutionary capabilities with making those capabilities feel accessible and relevant to specific customer problems rather than impressive but abstract technological achievements.

The Multiple Products Trap: Lessons from Rosebud

Rosebud's website exemplifies a common mistake among AI startups: trying to showcase multiple products simultaneously instead of focusing users on one primary value proposition and call-to-action.

• Competing messaging creates immediate confusion with "AI generated game assets," "game development," and "GPT to game" representing three different concepts presented simultaneously. Users can't quickly determine what the company actually does or offers.
• Multiple call-to-actions dilute focus with "gaming asset generator" and "AI Game Dev platform" buttons competing for attention alongside "try beta" options for different products. This creates decision paralysis rather than clear next steps.
• The four-product reveal continues scrolling with game assets, sprites, skyboxes, and avatar animation representing distinct offerings that should probably live on separate pages rather than competing for homepage real estate.
• Equal visual weight prevents users from understanding priority or recommended paths through the experience. When everything looks equally important, nothing feels important enough to click.
• Drop-off risk increases with every additional click required because some percentage of users abandon the process at each step. Making users choose between multiple options before taking action increases abandonment rates.
• Main KPI confusion emerges when companies can't clearly articulate their primary desired outcome from website visitors. Without clarity on the main goal, everything becomes secondary priority.

The solution involves identifying the primary product that drives business success and making that the clear focus of the homepage experience, with secondary products accessible through navigation but not competing for immediate attention.

Trust-Building Through Functional Design

AI startups must build trust rapidly because enterprise customers are skeptical of new technologies making bold claims about capabilities. Small design details significantly impact credibility perceptions.

• Interactive promise fulfillment becomes critical when websites claim "drag and drop" functionality but then fail to deliver working interactions. Broken expectations immediately damage trust and suggest the actual product may not work as advertised.
• Visual hierarchy precision matters more for enterprise AI tools because poor spacing and typography choices suggest lack of attention to detail that could extend to product quality. Small fit-and-finish issues create doubt about overall capabilities.
• Login wall placement affects user experience when "try it now" buttons lead directly to signup requirements without demonstrating value first. Users need to experience product benefits before committing to account creation.
• Animation purposefulness distinguishes between helpful motion that guides attention versus overwhelming sensory overload that distracts from core messaging. Every moving element should serve a specific communication goal.
• Social proof authenticity requires recognizable company names and specific testimonials rather than generic logos that could be outdated or exaggerated. Real people with faces and quotes feel more trustworthy than anonymous brand associations.
• Concrete claims with specificity like "30% cheaper and 25% faster" build more credibility than vague benefits, though sources and methodology questions arise when specific numbers appear without context or validation.

The underlying principle recognizes that AI startups operate in a trust deficit environment where potential customers have been oversold on AI capabilities, making credibility signals more important than flashy presentations.

Demonstrating AI Feats of Strength

AI products require different demonstration strategies than traditional software because the value comes from capabilities that seem magical or impossible to achieve through conventional means.

• Technology showcase videos should emphasize the "you've never been able to do this before" factor rather than generic product demonstrations. The wow factor drives initial interest and sharing behavior that builds awareness.
• Input-to-output magic represents the core AI value proposition where simple user inputs generate sophisticated results. Showing this transformation process in real-time creates the strongest impression of product capabilities.
• Before-and-after comparisons help users understand the time, effort, or skill savings that AI automation provides. The contrast between traditional approaches and AI-powered shortcuts demonstrates clear value.
• Real-time results demonstration builds confidence in product speed and reliability while showing actual capabilities rather than theoretical possibilities. Live demos carry more weight than prepared examples.
• Specific use case examples make abstract AI capabilities concrete by showing exactly how particular customers would use the product to solve actual problems they recognize and care about solving.
• Failure state acknowledgment builds trust by honestly showing what the AI can and cannot do rather than implying perfect results in all situations. Transparent limitations create realistic expectations.

The goal involves showing rather than telling how AI technology creates value, using concrete demonstrations that make abstract capabilities feel tangible and immediately useful for specific customer scenarios.

Specificity Over Generality in Messaging

AI startups often fall into the trap of trying to appeal to everyone with their revolutionary technology, but effective websites require clear focus on specific customer problems and use cases.

• Ideal customer profile clarity enables visitors to quickly self-identify whether the product is relevant to their situation. Generic messaging like "for any business" fails to create the personal connection needed for conversion.
• Problem-solution fit communication works better than technology-feature explanations because customers care about outcomes rather than implementation details. Leading with customer problems creates immediate relevance.
• Use case prioritization helps focus messaging on the most common and valuable applications rather than trying to cover every possible way the technology could be used. Depth beats breadth for initial user understanding.
• Customer language adoption means using the exact phrases that existing users employ to describe the product rather than internal technical terminology or marketing-invented descriptions.
• Competitor differentiation becomes essential in crowded AI markets where multiple companies claim similar capabilities. Clear explanation of unique advantages helps customers choose between options.
• Industry-specific examples resonate more strongly than generic demonstrations because they show understanding of particular customer contexts and challenges within specific business domains.

The principle recognizes that trying to be everything to everyone results in being nothing to anyone, particularly important for AI startups where the technology enables broad applications but marketing requires narrow focus.

Animation and Visual Design Best Practices

AI startup websites often overuse animation and visual effects, thinking that dynamic presentations better represent cutting-edge technology. However, purposeful design creates better user experiences than flashy animations.

• Selective animation usage means choosing specific elements to animate for attention direction rather than making everything move simultaneously. Too much motion creates sensory overload rather than engagement.
• Attention guidance through contrast helps users understand what actions to take next by making primary calls-to-action visually distinct from secondary options through color, size, and positioning choices.
• Information hierarchy establishment requires clear visual relationships between headlines, subheads, body text, and interactive elements so users can scan content efficiently and find relevant information quickly.
• Color psychology application involves understanding that bright, high-energy visuals may work for consumer products but enterprise customers often prefer more subdued, professional presentations that suggest reliability.
• Loading speed optimization becomes critical when animations and videos slow page performance, particularly important for AI websites that may already face skepticism about technical competence.
• Accessibility considerations ensure that motion doesn't interfere with screen readers or create problems for users with vestibular disorders who cannot tolerate excessive animation.

The key insight involves using design elements to support communication goals rather than impress visitors with technical prowess, recognizing that effective design feels invisible while poor design becomes a distraction.

Learning from Successful Examples: Voiceflow Analysis

Voiceflow demonstrates many best practices for AI startup websites, providing a positive example of how to effectively communicate complex technology capabilities to potential customers.

• Clear value proposition communication starts immediately with "teams use voice flow to design test and launch chat or voice assistants together faster at scale" - specific, benefit-focused, and immediately understandable.
• Effective demonstration strategy uses animated flowchart visualization that shows exactly what the product does without requiring technical explanation. Users can immediately grasp the concept through visual representation.
• Social proof integration throughout the page includes recognizable company logos, specific customer quotes with faces and names, and usage statistics ("100,000 teams") that build credibility without overwhelming the main message.
• Collaborative features highlighting addresses a key differentiator (multiple people working together) without getting lost in technical details or competing with the primary value proposition.
• Professional visual design creates trust through appropriate spacing, typography choices, and animation that guides rather than distracts from the core messaging.
• Established company advantage allows more confident, relaxed presentation because years of customer feedback and iteration have validated the approach, showing the value of learning from mature AI companies.

The success factors demonstrate how clarity, specificity, and user-focused design create more effective communication than flashy presentations or technical feature lists.

Garry Tan's website critiques reveal that AI startups face unique communication challenges requiring clear value propositions, focused messaging, and trust-building design elements. The most effective AI websites demonstrate capabilities through action rather than description, focus on specific customer problems rather than broad technology applications, and use design elements purposefully to guide user attention rather than impress with visual effects. Success comes from making revolutionary technology feel accessible and immediately useful for specific customer needs.

Practical Implications

• Lead with clear value propositions that immediately communicate what your AI product does and who it's for
• Focus on one primary call-to-action per page rather than offering multiple competing options that create decision paralysis
• Demonstrate AI capabilities through videos or interactive demos showing "feats of strength" rather than describing features
• Use customer language from actual user conversations rather than internal technical terminology or marketing jargon
• Implement social proof through recognizable company logos, specific testimonials with faces, and concrete usage statistics
• Design interactive elements that actually work as promised to build trust in overall product capabilities
• Apply animation sparingly and purposefully to direct attention rather than creating sensory overload with excessive motion
• Maintain professional visual hierarchy through proper spacing and typography to suggest enterprise-level quality
• Address common customer concerns proactively through FAQ sections and clear competitive differentiation
• Show specific use cases and customer examples rather than generic capabilities that could apply to anyone
• Build credibility through concrete claims with numbers while acknowledging limitations where appropriate
• Test all interactive elements to ensure they fulfill promises made in copy and avoid breaking user trust immediately