AI
Relic: From Multi-Agent Collaboration to Persistent Organizational Capability
Publication: Hugging Face Daily Papers
Author: Not available in collected metadata
Source published: 25/09/2026, 20:00:00 UTC
Briefing created: 29/09/2026, 22:13:31 UTC
Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 29/09/2026, 10:18:24 UTC.
Briefings
Relic transforms recurring collaboration failures into persistent organizational protocols by having members reflect on visible work, propose rules, and bind execution consequences to runtime. Across 360 controlled runs on ten workloads and three models, Relic improved complete-contract delivery from 14.06% to 19.76% (+5.71 percentage points). Behavioral correctness increased to 41.2% with executable bindings, a 6.5-point gain over text alone. On CooperBench, Relic achieved 76.9% accuracy, surpassing the best peer-structured system results.
Key points
- Relic converts collaboration friction into executable protocols that bind triggers, responsibilities, evidence, and consequences to runtime.
- Relic improved complete-contract delivery from 14.06% to 19.76% across 360 runs on ten workloads and three models.
Context and limitations
Results are from 360 controlled runs; full CooperBench results exclude broken benchmark pairs. Behavioral correctness data reflects fresh-member transfer scenarios only.
Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 29/09/2026, 22:13:31 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.
View archived briefing on IPFS
bafybeibgk23iruaymlkmpaqik2ptwfcmlrmofi4rey2qhlcvyxw7df34km