Been working on agent-swarm since Nov last year, and talking with a lot of companies on the exact same narrative that Claude Tag and Karpathy are pushing as "breaking change" and "the new UX era of AI".

We believe that as you move to an AI-native approach in your company you should be in control of the models, harnesses, context, and memory. It should not be a blackbox as Claude is doing.

Hence the swarm is trying to solve this problem: and open project that let you define a team of agents that meet you where you work. Just @agent-swarm in any channel and you good to go. Want to continue the convo in email? You can.

It is FOSS and we aim to always be. AI models might be closed, but how you integrate it should be open.

https://github.com/desplega-ai/agent-swarm

Cheers,