It depends. I am a data scientist. Below is why its recently started to work for me:
when I have for example 5 streams of experiments that I am working for months, with some loose connections, I prefer having each as an "agent" with sort of long memory. They can talk with each other but still have some independency. So each agent remembers its previous tasks very well but also at some level familiar with other agents work.
To manage all these individual agents, I have one TPM agent that I actually talk to. And it handles my tasks by communicating with those individual agents.
Wondering what you think? Is it better ways for my situation and use case
It depends. I am a data scientist. Below is why its recently started to work for me:
when I have for example 5 streams of experiments that I am working for months, with some loose connections, I prefer having each as an "agent" with sort of long memory. They can talk with each other but still have some independency. So each agent remembers its previous tasks very well but also at some level familiar with other agents work.
To manage all these individual agents, I have one TPM agent that I actually talk to. And it handles my tasks by communicating with those individual agents.
Wondering what you think? Is it better ways for my situation and use case
[flagged]