A Beginner-Friendly Tutorial on LLM-based Agents
This tutorial is designed for researchers and engineers who are interested in large language model (LLM)-based agents but have little or no prior experience with LLMs, machine learning, artificial intelligence, or programming. It offers a gentle yet comprehensive introduction, providing newcomers with an intuitive, high-level understanding of what LLM agents can do, while also serving as a practical guide for those who wish to explore how these agents actually work. Key takeaways from the tutorial include:
- foundational concepts of LLMs and LLM-based agents;
- a curated set of nine concrete examples demonstrating essential agent capabilities and the mechanisms behind them across diverse domains, including scientific discovery, social simulation, software engineering, finance, biology, and medicine;
- practical tips for developing LLM agents;
- a brief roadmap of advanced techniques in this rapidly evolving field, such as injecting domain knowledge, improving reasoning abilities, and managing long context.
From foundations to frontier
The diagram below mirrors that structure so you can jump directly to the topic you need.
1. What you must know
Preliminaries on next-token prediction, context windows, and six hallmark agent capabilities.
2. What agents can do
Nine representative agents spanning social simulation, science, biology, medicine, software, finance, gaming, mobile devices, and robotics.
3. Practical tips
Keep designs simple, define goals clearly, guide with human workflows, and rely on example-driven prompts.
4. Advanced frontier
Inject domain knowledge, strengthen reasoning, manage long context, internalize agent skills, and keep agents safe.
If you find this work helpful, please consider citing our paper:
@article{hu2025llm_agents_tutorial,
title={A Beginner-Friendly Tutorial on LLM-based Agents},
author={Hu, Shuyue and Ren, Siyue and Chen, Yang and Mu, Chunjiang and Liu, Jinyi and Cui, Zhiyao and Zhang, Yiqun and Li, Hao and Zhou, Dongzhan and Xu, Jia and Zhang, Qiaosheng and Han, Chu and Zheng, Yan and Hao, Jianye and Wang, Zhen},
year={2025},
month={November},
note={Manuscript in preparation}
}