Hello everyone, Im trying to replicate the work of a paper called ProtAgents which uses the framework autogen to create a multiagent system to design proteins. I was wondering if changing to langgraph would improve it or will it add unnecessary work and complexity?
I think langgraph is more easy to use then autogen as the later involves more customisation of built-in prompt allowing users to have flexibility when creating an agent, where as langgraph will basically use langchain to create your agent based on probabilities and represent as a graph.
From what I have known for complex application, autogen is better choice due to its flexibility of converting agents into conversational framework allowing the creator to create its own kind of code or prompt based on their requirements.