The latest release of the assistant API makes it simple to experiment directly within the default OpenAI playground, but I want to push it further, so I want to have a different kind of playground that is tailored for experimenting with multiple assistants/agents swarm.
After talking to a few people, here is what I have in mind for this supposedly “pro” playground:
- conversating with multiple assistants, so no switch tab back and forth
- orchestrate agents swarm, and have a high-level overview of what’s going on (watching the assistants converse with each other)
- visualize the conversation flow, and the agents’ state
What do you think? What would you like to see in this playground? What would you like to experiment with?
Here is the progress so far: