Feed
researchAI Agent Research11d ago

Researchers Introduce Evolver Framework for Persistent AI Agent Experience

Facts

Evolver is an experience-driven framework that captures agent execution traces as reusable 'Gene' assets, enabling capability evolution without retraining foundation models. Empirical testing demonstrates significant improvements in task success rates and efficiency across multiple agent benchmarks.

  • ▸Evolver stores procedural knowledge as structured Gene assets for reuse in future agent tasks.
  • ▸The framework operates without modifying foundation model parameters.
  • ▸Tested with Claude Code and Codex configurations across four benchmarks including SWE-bench and GAIA.
  • ▸Task success improved by 4.38 percentage points on a 320-execution diagnostic set.
  • ▸Recorded task-execution token consumption reduced by 24.24%.
Occurred
2026-11-05 07:44 UTC
First reported
2026-11-05 07:44 UTC
Last updated
2026-11-06 04:10 UTC
Event type
research
Region
Global

Entities

Categories

Source timeline · 1

  • arXivFirst-party2026-11-05 07:44
Community vote

Sign in to vote on importance and sentiment.

Platform Signals

Impact score
69
AI sentiment
Positive
Event significance30%1
Entity importance20%1
Source credibility20%1
Novelty15%1
Corroboration15%1
Confidence1
Novelty1
Corroboration1
Community
ImportanceNo votes
Strategic sentiment
Positive 0%Neutral 0%Negative 0%

Platform signals are algorithmic and community-derived indicators. They are not investment recommendations and do not predict valuation or performance.