4,493 followers
Big news: Andi ranks #1 on a new SOTA AI search benchmark! 🚀 Andi achieved 87% accuracy in the new SearchBench AI evaluation by Talc AI (YC S23), outperforming tech giants and well-funded startups alike. Google Gemini scored 71%. And Perplexity just 59%. Here's why this matters: 1. Real-world accuracy: SearchBench generates 900 questions in real-time based on live data and current events. Aims to measure real-world searches, and resist attempts to 'game' evaluations. 2. David vs. Goliath: We did this with only $3M in funding, compared to hundreds of millions or even billions for our competitors. 3. User-focused innovation: We're obsessed with understanding content deeply, not just matching keywords. Our Trantora backend analyzes credibility and relevance, delivering trustworthy results. Here's a real example measured in the test set: Q: "First female artist album with five Billboard Hot 100 number-one singles?" ✅ Andi: Correctly answered Katy Perry's "Teenage Dream" ❌ Gemini: Incorrectly said Taylor Swift's "1989" ❌ Perplexity: Contradicted itself, mentioning Madonna and Mariah Carey. This shows how Andi integrates multiple pieces of information that would typically require several Google searches. We're building a search experience focused on what users actually need. We're just getting started. Our goal isn't only to change search - we're building a trust engine for the AI age. Try Andi for yourself at https://andisearch.com and experience the most accurate AI search engine available today! Curious about the details? Check out the full benchmark results from Talc here: https://lnkd.in/gM3mFWXb We'd love to hear your thoughts. How do you think this kind of accuracy will impact the future of search and AI? 😀 #AI #Search #Innovation #YCombinator #TechStartup
