here's how my solo ai company works: we have departments like planning ⚡coding⚡ review. sounds normal right?
but then there's this one thing that sets us apart -
who improves the agent after it makes mistakes usually, you just tweak settings or restart but in our case. someone actually sits down and rewrites parts of how an ai does its job.
this 'improver' role is key because once they start refining agents based on past errors the whole system starts to evolve faster. it's like having a team that learns from each other, not just one-off fixes
anyone else tried something similar? i'd love some feedback or if anyone has cool tips for making ai more self-improving!
article:
https://dev.to/setas/the-improver-how-i-built-an-ai-agent-that-upgrades-other-ai-agents-2l9j