Settings

Theme

Show HN: Agent Actors – Plan-Do-Check-Adjust with Parallelized LLM Agent Trees

github.com

8 points by knrz 3 years ago · 3 comments · 1 min read

Reader

Hey HN! Been working on this library for architecting stateful LLM agent trees that execute in parallel. Think of it like a AI scheduler for BabyAGIs or AutoGPTs, like:

-----

Parent (Plan and Adjust): Chief Revenue Officer / VP Sales

Children (Do and Check): 3 Sales Development Representatives; 2 Account Managers; 1 Market Researcher.

You can give the CRO a task and it will break it down, distribute it appropriately to its children, and the children will work in parallel on the task.

-----

Curious to hear your feedback first HN, we're launching on Twitter tomorrow!

mitko 3 years ago

The VP of Sales is interesting. I would imagine that agents wouldn't look like the traditional human roles, but would stack fractally under each other, like functions in code do.

  • knrzOP 3 years ago

    I found modelling them as humans made it easier to imagine how to structure them, but there's nothing saying that's it's a better way to model your agent tree. Would be curious for somebody to use this to run some experiments with different names/traits for the fractal agent tree.

olahmoon 3 years ago

This is brilliant and incredibly useful. Also easy to customize.

Keyboard Shortcuts

j
Next item
k
Previous item
o / Enter
Open selected item
?
Show this help
Esc
Close modal / clear selection