The MCP Server tools are quite limited compared to what AI Teammates can do. This limits what you can do in (or based on data in) Asana with external AIs. I appreciate there is probably a commercial reason for this, but I don’t think it is good UX.
I’m not certain it makes sense for there to be full parity between the two, but closing the gap would make a lot of workflows much easier. Tools like ChatGPT and Claude are where many people centralise their workflows now. These tools have much broader integrations with most people’s entire tech stacks than it is probably feasible for AI Teammates to ever have. Being able to have ChatGPT/Claude fully understand my Asana to orchestrate action in Asana (and elsewhere) is a far smoother workflow than trying to trigger everything from Asana AI Teammates.
In theory, AI Teammates are using primitives that could probably be exposed via the MCP Server to allow external AIs to get closer to AI Teammate functionality. I don’t know how complex it would be to code, but it doesn’t seem outlandish.