Why is this so simple?
Couldn’t OpenAI, such a large company, create a more advanced unit framework? Even amateur developers have better projects.
That would compete with themselves now wouldn't it?!
On Wed, Jul 30, 2025 at 8:46 AM tarikkaya @.***> wrote:
tarikkaya created an issue (openai/swarm#59) https://github.com/openai/swarm/issues/59
Couldn’t OpenAI, such a large company, create a more advanced unit framework? Even amateur developers have better projects.
— Reply to this email directly, view it on GitHub https://github.com/openai/swarm/issues/59, or unsubscribe https://github.com/notifications/unsubscribe-auth/BA5P2ADWGHLX2FIFQC7OUW33LDSGBAVCNFSM6AAAAACCXKZCF6VHI2DSMVQWIX3LMV43ASLTON2WKOZTGI3TONJZGMZTINY . You are receiving this because you are subscribed to this thread.Message ID: @.***>
That would compete with themselves now wouldn't it?! …
Is this an open AI vs. closed AI debate? No, I don’t think so.
What is the discussion all about?
@tarikkaya Take a look at agent-swarm-kit. It based on openai swarm architecture pattern
@tarikkaya Take a look at agent-swarm-kit. It based on openai swarm architecture pattern
I’ve looked at that project and other similar ones. The unit logic is very simple. Yet you could feed each unit with both prompt and training vectors, as well as a standard database and a Skala DB. The databases could support both local and cloud providers. It’s all been implemented in a very simple, amateurish way. Units don’t have to focus on a single domain—they can cover multiple fields of work. They can be used to master multiple lines of thought and perspectives. I’m working on this; I don’t know if I can do it alone. But I know that as a team, an excellent job can be done.
Your flock’s wings are missing.
Why shouldn’t units have training and experience vectors in the work they do? If you create a unit on a topic, why shouldn’t that unit specialize in that topic?