Source URL: https://simonwillison.net/2025/Jan/11/agents/
Source: Simon Willison’s Weblog
Title: Agents
Feedly Summary: Agents
Chip Huyen’s 8,000 word practical guide to building useful LLM-driven workflows that take advantage of tools.
Chip starts by providing a definition of “agents" to be used in the piece – in this case it’s LLM systems that plan an approach and then run tools in a loop until a goal is achieved. I like how she ties it back to the classic Norvig "thermostat" model – where an agent is "anything that can perceive its environment and act upon that environment" – by classifying tools as read-only actions (sensors) and write actions (actuators).
There’s a lot of great advice in this piece. The section on planning is particularly strong, showing a system prompt with embedded examples and offering these tips on improving the planning process:
Write a better system prompt with more examples.
Give better descriptions of the tools and their parameters so that the model understands them better.
Rewrite the functions themselves to make them simpler, such as refactoring a complex function into two simpler functions.
Use a stronger model. In general, stronger models are better at planning.
The article is adapted from Chip’s brand new O’Reilly book AI Engineering. I think this is an excellent advertisement for the book itself.
Via @chiphuyen.bsky.social
Tags: ai-agents, llms, ai, generative-ai, llm-tool-use
AI Summary and Description: Yes
**Summary:** The text discusses Chip Huyen’s comprehensive guide on building workflows with LLM-driven agents, exploring concepts like planning within AI systems and offering practical advice for better tool integration. This content is particularly relevant for professionals in AI and LLM security, as it touches on effective methodologies for ensuring that AI agents function optimally and securely.
**Detailed Description:** This text presents a summary of Chip Huyen’s extensive 8,000-word guide focused on the development of workflows using large language models (LLMs) in the context of AI agents.
Key insights include:
– **Definition of Agents:**
– Agents are described as LLM systems capable of recognizing their environment (perception) and taking actions (acting).
– Huyen parallels this with the classic thermostat model by Norvig, which underscores the capability of agents to perceive (via read-only actions or sensors) and act (via write actions or actuators).
– **Planning in AI Systems:**
– The article emphasizes planning as a critical aspect of AI agents, offering substantial guidance on enhancing the planning process:
– **Improving System Prompts:** Crafting better system prompts by incorporating more examples to drive clearer understanding.
– **Tool Descriptions:** Providing detailed descriptions of the tools and their parameters, enabling the AI to understand its resources effectively.
– **Function Simplification:** Encouraging the rewriting of complex functions into simpler components to streamline the process.
– **Using Stronger Models:** Highlighting that leveraging more sophisticated models generally leads to improved planning capabilities.
– **Promotion of AI Engineering Book:**
– This piece is an adaptation from Huyen’s newly released O’Reilly book, indicating its role as both a practical guide and a promotional piece for his written work.
These insights not only potentiate efficiency in constructing AI solutions but also underscore the importance of meticulous planning and execution in LLM applications. For professionals, especially those in AI security, recognizing the nuances of agent behavior and tool utilization is crucial in building secure and robust AI infrastructures.
Overall, the article is a strong resource for those interested in the advancement of AI engineering, particularly in the application of LLMs and the ongoing discourse on AI’s role in security and infrastructure.