AI Summary
→ WHAT IT COVERS Cursor co-founder Sualeh Asif explains building the popular AI coding editor, scaling to 100 million daily requests, infrastructure challenges, and the future of AI-assisted programming. → KEY INSIGHTS - **Model Selection:** Cursor uses DeepSeek models for their tab completion feature, processing hundreds of millions of requests daily due to superior pre-training and cost efficiency for coding tasks. - **Product Philosophy:** Ship features only when they feel genuinely useful in daily coding - if developers don't use it themselves for 10 hours daily, users won't adopt it either. - **Infrastructure Strategy:** Build custom indexing infrastructure handling billions of files daily using disaggregated storage on S3 with write-ahead logs for vector database operations at massive scale. - **Context Windows:** Large context windows (50,000-60,000 tokens per request) became crucial for agent functionality - models can't help effectively if they can't read your current file. → NOTABLE MOMENT Asif reveals Cursor was originally named "Copilot Plus Plus" internally, and the AI model still needs reminders about this naming convention when searching their own codebase. 💼 SPONSORS None detected 🏷️ AI Coding Tools, Software Development, Machine Learning Infrastructure, Developer Productivity