// AI news · 2026-11-10 · 量子位
openJiuwen Launches TTSE Dual-Track Self-Evolution Framework
openJiuwen introduces TTSE, a dual-track self-evolution framework that separates environmental understanding from procedural knowledge in agent learning.
Published 2026-11-10 15:56 (Beijing time)Source 量子位Auto-summarised, fact-checked against source
01Summary
The system continuously validates and updates facts about the environment and action methods through task feedback loops. Experiments across multiple benchmarks show consistent performance gains over single-track approaches, with notable improvements in success rates and efficiency in real-world office tasks.
Summarised from the original article by an open-source model, then checked automatically: every number and English name must appear in the source. The original article is authoritative.
02Key points
- TTSE splits experience into environmental facts (FACT) and conditional procedures (TIP).
- Uses a cycle of blame, retirement, contradiction analysis, and induction to iteratively refine knowledge.
- Outperforms single-track methods across ALFWorld, ScienceWorld, GDPevo, SOPBench, and PinchBench benchmarks.
- Integrated into WorkSwarm 0.2.7+; users can enable TTSE to observe efficiency gains in real office tasks.
03Source
Agent自演进驶入“双轨时代”!openJiuwen首发TTSE经验学习框架
量子位 · published 2026-11-10 15:56 Beijing time
Also reported by 1 other outlet
05Related news
- Same source · Nvidia Releases GR00T-2 Robot Foundation Model2026-11-10 · 量子位
- Same source · AI Disrupts University CS Education2026-11-09 · 量子位
- Same topic · UK Government Sites Reveal Sensitive Data2026-11-10 · WIRED
- Same topic · Google Launches Gemini for Healthcare2026-11-10 · Google Cloud AI/ML
- Same day · Jensen Huang withdraws from events after admitting extramarital affair2026-11-10 · IT之家
From AIMap AI news: collected from 1,100+ official and media sources. Summaries only restate the original article; follow the link for the full story. All stories of 2026-11-10 · Archive by date · All AI news · Today