LEGOMem: Modular Procedural Memory for Multi-agent LLM Systems for Workflow Automation
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912631478026240 |
|---|---|
| author | Han, Dongge Couturier, Camille Diaz, Daniel Madrigal Zhang, Xuchao Rühle, Victor Rajmohan, Saravan |
| author_facet | Han, Dongge Couturier, Camille Diaz, Daniel Madrigal Zhang, Xuchao Rühle, Victor Rajmohan, Saravan |
| contents | We introduce LEGOMem, a modular procedural memory framework for multi-agent large language model (LLM) systems in workflow automation. LEGOMem decomposes past task trajectories into reusable memory units and flexibly allocates them across orchestrators and task agents to support planning and execution. To explore the design space of memory in multi-agent systems, we use LEGOMem as a lens and conduct a systematic study of procedural memory in multi-agent systems, examining where memory should be placed, how it should be retrieved, and which agents benefit most. Experiments on the OfficeBench benchmark show that orchestrator memory is critical for effective task decomposition and delegation, while fine-grained agent memory improves execution accuracy. We find that even teams composed of smaller language models can benefit substantially from procedural memory, narrowing the performance gap with stronger agents by leveraging prior execution traces for more accurate planning and tool use. These results position LEGOMem as both a practical framework for memory-augmented agent systems and a research tool for understanding memory design in multi-agent workflow automation. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_04851 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | LEGOMem: Modular Procedural Memory for Multi-agent LLM Systems for Workflow Automation Han, Dongge Couturier, Camille Diaz, Daniel Madrigal Zhang, Xuchao Rühle, Victor Rajmohan, Saravan Artificial Intelligence Machine Learning Multiagent Systems We introduce LEGOMem, a modular procedural memory framework for multi-agent large language model (LLM) systems in workflow automation. LEGOMem decomposes past task trajectories into reusable memory units and flexibly allocates them across orchestrators and task agents to support planning and execution. To explore the design space of memory in multi-agent systems, we use LEGOMem as a lens and conduct a systematic study of procedural memory in multi-agent systems, examining where memory should be placed, how it should be retrieved, and which agents benefit most. Experiments on the OfficeBench benchmark show that orchestrator memory is critical for effective task decomposition and delegation, while fine-grained agent memory improves execution accuracy. We find that even teams composed of smaller language models can benefit substantially from procedural memory, narrowing the performance gap with stronger agents by leveraging prior execution traces for more accurate planning and tool use. These results position LEGOMem as both a practical framework for memory-augmented agent systems and a research tool for understanding memory design in multi-agent workflow automation. |
| title | LEGOMem: Modular Procedural Memory for Multi-agent LLM Systems for Workflow Automation |
| topic | Artificial Intelligence Machine Learning Multiagent Systems |
| url | https://arxiv.org/abs/2510.04851 |