Qwen's Personal Agent deep dive: 8,000 atomic capabilities, memory design and 40% context reduction
机器之心 · wechat · 2026-11-05
Machine Heart interviews Jiang Guanjun, AI algorithm lead of Alibaba's Qwen unit, on the technical framework behind the invite-testing Personal Agent "Qwen Go!".
Three pillars: model, memory, ecosystem. Qwen built demand maps across 30+ industries, decomposed them into 8,000+ atomic model capabilities, and uses post-training/RL to improve task stability. A flagship case: the agent persisted on a user's cash-compensation goal during a flight-delay claim, rejecting voucher offers and securing ¥70 cash.
Memory: demand-driven schema—decide what to store based on how it will be used; recall by event association and causality, with update/expiration mechanisms.
Engineering: model routing, skill layering, and context compression (history distilled into memory) cut context size by 40%. Proactive service is early-stage; the long-term bet is messy life scenarios like home buying and schooling.
More from coding & agent
- Gemini /teamwork agents caught live: four parallel task logs all running that second — RyanSpunk · 2026-11-13
- Codex CLI 0.168.0 adds model provider picker, side-by-side TUI workspace and animated GIF generation — github-actions[bot] · 2026-11-13
- Free MCP server lets Claude read Altium, Allegro, KiCad and Gerber PCB files directly — woodisruff · 2026-11-13
- a16z Partner Clones Open-Source Game with Fable 5.2 Agent, Which Finds Its Own Horse Animation Asset — FinanceYF5 · 2026-11-13
- OpenAI team shares how it runs and governs fleets of persistent agents internally — romainhuet · 2026-11-13
- Perplexity's Decider scanned 1B tokens across 7,000 agent sessions, found 14 security fixes — denisyarats · 2026-11-13