bguberfain on LLM function calls don't scale; code orchestration is simpler, more effective

Comment on LLM function calls don't scale; code orchestration is simpler, more effective

I think that there may be another solution for this, that is the LLM write a valid code that calls the MCP's as functions. See it like a Python script, where each MCP is mapped to a function. A simple example:

  def process(param1, param2):
     my_data = mcp_get_data(param1)
     sorted_data = mcp_sort(my_data, by=param2)
     return sorted_data

jngiam1OP1y

Yes! If you want to see how this can work in practice, check out https://lutra.ai ; we've been using a similar pattern there. The challenge is making the code runtime work well for it.

Comments