DeepSeek's Super-agent blueprint meets AFKAI's Comma: one session, router plus many workers
量子位 · wechat · 2026-11-07
After DeepSeek's Harness team published its Super-agent vision, the Silicon Valley Chinese team AFKAI's product Comma drew attention for converging on the same architecture months earlier. Shared diagnosis: humans can only track 3–5 active sessions while agent concurrency scales unbounded. Shared design: a single persistent conversation where a Router orchestrates context-isolated Workers in parallel — continuous, efficient, proactive, persistent. Comma specifics: cross-app execution via its Salix runtime (WeChat, Signal, iMessage, Linear; Linear issues can flow back to Comma), 24/7 proactive watch even when the client is closed, long-term memory kept outside model context, and continual learning that distills human corrections into small models upgraded monthly. Timeline: Bridge (predecessor) launched April 28, 2026 with LeadAgent + Subagents; open-sourced Sept 30, with Lean-formalized safety mechanisms. Anthropic's similar Claude Tag arrived 20 days after the fixed-router mechanism.
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