Meta's Muse hits ~9M downloads; Alexandr Wang details the 2-week prototype and 7-month grind
快鲤鱼 · wechat · 2026-10-15
A translated deep-dive of David Senra's interview with Meta Chief AI Officer Alexandr Wang on how the personal AI agent Muse was built. Launched Sept 8, Muse can handle email, travel booking, forms, and transactions; Sensor Tower data shows 3.4M+ downloads by late September (#1 on the US App Store) and 9M cumulative by October.
- Origin: in February, Nat Friedman and Wang went deep on OpenClaw and wrote memos declaring 'super-agents may be the ultimate consumer product'
- A 1-2 week prototype, then 7 months of reliability grinding: a 100+ item capability checklist, shipped only when MuseSpark 1.3 passed every gate; team under 200 people
- Strategy: while the industry chased coding agents, Meta bet on consumer personal agents, leveraging its 3.5B-user app ecosystem
- Growth: Wang's own meme posts and real-user screenshots drove organic spread; retention comes from a 'trust fall' loop of escalating delegation
Context: Meta paid $14.3B for 49% of Scale AI ($29B valuation) in June 2025, bringing Wang in to lead its superintelligence lab.
More from Companies & People
- Humanities dissertations have collapsed, social sciences are next — with AI as one driver — gsiemens · 2026-10-23
- For consumer AI founders, 'When are you launching in Europe?' is the new 'When Android?'—only 100x worse — omooretweets · 2026-10-23
- Lilian Weng to Discuss Recursive Self-Improvement at SGLang Summit 2026 — BanghuaZ · 2026-10-23
- OpenAI FDE team: make custom work obsolete by feeding customer learnings into products — colintjarvis · 2026-10-23
- Bindu Reddy: record AI security incidents, tech stock slide, rogue agents rattle banks — bindureddy · 2026-10-23
- StepFun shows up at AGNTCon and MCP Con North America with booth #24 — StepFun_ai · 2026-10-23