# Agent Design Pattern Catalogue: A Collection of Architectural Patterns for Foundation Model based Agents

**Authors:** Yue Liu, Sin Kit Lo, Qinghua Lu, Liming Zhu, Dehai Zhao, Xiwei Xu, Stefan Harrer, Jon Whittle  
**arXiv:** [2405.10467](https://arxiv.org/abs/2405.10467)  
**Date:** 2024-05-16  
**Summary page:** https://www.summarizepaper.com/en/paper/2405.10467v4/

## Key points

- The paper discusses the use of foundation model-enabled generative artificial intelligence to develop agents with advanced reasoning and language processing capabilities.
- A systematic literature review was conducted to identify 18 architectural patterns that can address challenges such as generating instrumental goals and plans, dealing with hallucinations, ensuring explainability of reasoning process, and managing complex accountability.
- The authors propose a decision model for effectively selecting these architectural patterns.
- The pattern catalogue serves as a guide for practitioners designing foundation model-based agents by facilitating goal-seeking and plan generation.
- The research aims to support architects in creating agents that can navigate goal pursuit complexities while leveraging generative artificial intelligence capabilities.

## Summary

The paper "Agent Design Pattern Catalogue: A Collection of Architectural Patterns for Foundation Model-based Agents" by Yue Liu et al. delves into the use of foundation model-enabled generative artificial intelligence to develop and implement agents with advanced reasoning and language processing capabilities. The authors conducted a systematic literature review to understand the current state-of-the-art in this field and identified 18 architectural patterns that can help address challenges such as generating instrumental goals and plans, dealing with hallucinations inherent in foundation models, ensuring explainability of the reasoning process, and managing complex accountability. They also propose a decision model for effectively selecting these patterns. This pattern catalogue serves as a comprehensive guide for practitioners looking to design foundation model-based agents by facilitating goal-seeking and plan generation. Through their research on effective use of architectural patterns, the authors aim to support architects in creating agents that can successfully navigate the complexities of goal pursuit while leveraging the capabilities of generative artificial intelligence.