In practice, you can build tools for whatever you want (at the end of the day they are just functions the LLM can use), from a tool that let's you acc

Search code, repositories, users, issues, pull requests...

submited by

Style Pass

2024-09-30 11:00:02

In practice, you can build tools for whatever you want (at the end of the day they are just functions the LLM can use), from a tool that let's you access Wikipedia, another to analyse the content of YouTube videos or calculate difficult integrals in Wolfram Alpha.

So, we've seen agents capable of reflecting and using tools to access the outside world. But ... what about planning, i.e. deciding what sequence of steps to follow to accomplish a large task?

That is exactly what the Plannin Pattern provided; ways for the LLM to break a task into smaller, more easily accomplished subgoals without losing track of the end goal.

These frameworks implement different variations of the multi-agent pattern, in which tasks are divided into smaller subtasks executed by different roles (e.g. one agent can be a software engineer, another a project manager, etc.)