Research paper

RAFT: A Stateful Retrieval-Augmented Framework for Troubleshooting Agents

Research from MicrosoftPublished by Association for Computational LinguisticsOct 2026PDF

Learn how retaining troubleshooting state can help an assistant select relevant evidence as a problem evolves.

RAFT: A Stateful Retrieval-Augmented Framework for Troubleshooting Agents cover

Resource access

Get the full resource

Enter your business details to read the full resource.

First nameLast nameWork emailCompanyJob titleCountry or region

Mingxuan Zhang, Xiaowen Wang, Anupma Sharan, Zhengyi Chen, Chenyu Diana Zhang, Shanshan Yang, Chittibabu Pacharu. RAFT: A Stateful Retrieval-Augmented Framework for Troubleshooting Agents. Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing: Industry Track. © 2026 Association for Computational Linguistics. CC BY 4.0. 10.18653/2026.emnlp-industry.21. Original PDF, unmodified.

CC BY 4.0

RAFT: A Stateful Retrieval-Augmented Framework for Troubleshooting Agents

Related research.

Cache-aside caching using Azure Managed Redis and Azure Database for PostgreSQL

Event-driven cache updates with Azure Managed Redis and Azure Cosmos DB

Azure NetApp Files Object REST API for data and AI integration