# Researchers report stronger agent training by adding guidance to practice tasks

_Published Friday, September 25, 2026 at 3:07 PM EDT · AI, Science · Latest · Tier 2 — Notable_

Researchers report that their training method improved AI agents' performance on two multi-step task benchmarks. Across three student models, the method produced the best average score in every tested setting, with gains of up to 65% in task-goal completion on AppWorld and up to 61% in resolved rate on SWE-bench Verified.

The method gives an agent a short, task-specific instruction when its attempts mostly fail, then trains on another attempt. The authors say a competing technique can teach agents to act as though they have information they never received. These results come from a preprint and its reported benchmark tests.

## Sources

- [cs.AI updates on arXiv.org](https://arxiv.org/abs/2609.29051)

---
Canonical: https://techandbusiness.org/newswire/wRTIaHgfWjsjvtjNtDU_bi
Published: 2026-09-25T19:07:40.953Z
Story chronology: 2026-09-25T04:00:00.000Z
Retrieved: 2026-09-25T22:18:54.766Z
Publisher: Tech & Business (techandbusiness.org)