The Subconscious agent is running well, and it's so fun to watch.
I made my research pipeline the target of improvement for the subconscious to ideate an improvement.
Outcome:
- Accepted during synthesis
- Confidence: medium
- LLM calls: 13
- Cost: $0.0739
A few tweaks before I share this publically:
- I added a synthesis layer during the debate, so I need to make sure ideas aren't easily accepted during this phase
- Experiment with increased idea amount (currently 5)
- Remove restrictions of ideas to solely its own context
For a v1, these fixes should be simple with the help of Codex and Claude.
Turn on notifications for when the full guide drops later today.
Graeme
@gkisokay
04-02
The #1 criticism I've received about the self-improving multi-agent framework is that you can't control the agents' outputs.
So I built a “Subconscious agent.”
Inspired by @karpathy’s autoresearch, it’s an LLM process that continuously looks for useful problems to solve.
All x.com/gkisokay/statu…
Join The Agent Network, my free AI builder community.
The waiting room is open.
From Twitter
Disclaimer: The content above is only the author's opinion which does not represent any position of Followin, and is not intended as, and shall not be understood or construed as, investment advice from Followin.
Like
Add to Favorites
Comments
Share
Relevant content
