# Researchers introduce ReScopeBench to test long-lived AI agents’ experience retention

Digest AI · Agents & Tools · published 2026-11-05T05:00:00Z

Canonical: https://digestai.news/story/researchers-introduce-rescopebench-to-test-long-lived-ai-agents-experi

## Summary

A new benchmark called ReScopeBench evaluates how AI agents manage experience over time in changing environments. The paper, posted on arXiv, focuses on agents that assist users across multiple tasks rather than single interactions. It distinguishes between structural knowledge (how an environment works) and rule-based knowledge (principles guiding decisions).

The authors propose an experience-first approach to build task streams around shared learning, then track how agents retain, revise, or lose experience. Across 20 models, they find retained experience boosts performance but agents fail to update it when environments change. Even with correct experience, tasks often remain unsolved, highlighting gaps in selective maintenance and effective use.

## Key points

- ReScopeBench tests 18 interactive environments to measure how agents retain, revise, or lose experience
- 20 models tested show retained experience improves task performance but fails to adapt to changes
- Researchers separate structural knowledge (environment organization) from rule-based knowledge (task principles)

## Why it matters

Understanding how agents manage experience over time is critical for building reliable, long-term AI assistants. This work pinpoints where current systems fail and could guide future improvements in adaptability and memory retention.

## Sources

1. [Towards Robust Long-Lived Agents: Benchmarking Experience Lifecycles in Evolving Environments](https://arxiv.org/abs/2611.05450) (arXiv cs.AI, 2026-11-05, primary source)

## Cite

Digest AI, "Researchers introduce ReScopeBench to test long-lived AI agents’ experience retention", 5 November 2026, https://digestai.news/story/researchers-introduce-rescopebench-to-test-long-lived-ai-agents-experi

---

Written by Digest AI's editorial model from the linked sources; the sources are the record. Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse
JSON: https://digestai.news/story/researchers-introduce-rescopebench-to-test-long-lived-ai-agents-experi.json