Federation over Text: Insight Sharing for Multi-Agent Reasoning

2026-04-17

AI Summary

This work addresses the absence of cross-agent knowledge sharing mechanisms in existing large language model agents when tackling novel tasks. It proposes FoT, a semantic-level federated reasoning framework that, for the first time, enables multiple agents to iteratively share and aggregate their local reasoning trajectories without gradient updates or supervisory signals, thereby constructing a transferable and reusable metacognitive insight repository. This approach facilitates semantic knowledge federation across tasks and domains, significantly enhancing reasoning efficiency and performance: it achieves an average accuracy improvement of 24% and reduces reasoning token consumption by 28% on mathematical problem-solving and cross-domain collaborative tasks, while capturing over 90% of the core contributions found in subsequent research papers in scientific insight discovery scenarios.

Technology Category

• Multiagent Systems: Adversarial Agents
• Knowledge Representation and Reasoning: Common-Sense Reasoning
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Agentic search
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

LLM-powered agents often reason from scratch when presented with a new problem instance and lack automatic mechanisms to transfer learned skills to other agents. We propose a federated learning-like framework, Federation over Text (FoT), that enables multiple agents solving different tasks to collectively generate a shared library of metacognitive insights by iteratively federating their local reasoning processes. Instead of federation over gradients (e.g., as in distributed training), FoT operates at the semantic level without any gradient optimization or supervision signal. Iteratively, each agent does local thinking and self-improvement on their specific tasks independently, and shares reasoning traces with a central server, which aggregates and distills them into a cross-task (and cross-domain) insight library that existing and future agents can leverage to improve performance on related tasks. Experiments show that FoT improves reasoning effectiveness and efficiency across a wide range of challenging applications, including mathematical problem solving, cross-domain collaboration, and machine learning research insight discovery. Specifically, it improves average accuracies of downstream tasks by 24% while reducing the reasoning tokens by 28% across the first two applications. In the research insight discovery application, FoT is able to generate insights that cover over 90% of the major contributions in the subsequent papers.

Problem

• multi-agent reasoning
• knowledge sharing
• metacognitive insights
• federated learning
• cross-task transfer

Innovation

• Federation over Text
• metacognitive insights
• multi-agent reasoning
• semantic-level federation
• LLM agents