Curious about external AI model connection

This is more a curiosity question than anything. I recently interviewed with a potential new customer, and one of their technical staff cited the inability to connect to an AI model through LangChain as a shortcoming of A360. Today apparently they are doing this through python. I am no expert in LangChain, but I would think this could be done through a task calling python code and configuring the connection to the model. You would of course need the correct libraries installed on the Bot Runner(s). From there you could call the appropriate LangChain tools as needed.

Am I way off on this? When I suggested this method the customer’s technical engineer scoffed like I am crazy.