#Which llm follows tool calling
1 messages · Page 1 of 1 (latest)
I have loosely similar setup. I do not populate the form in realtime, I collect all the information in a back and forth conversation and send all data to the form at once at the end (Custom MCP server). Of course, I have no idea if this works for your use case, but that's what we're doing.
Or maybe you could do it in batches to help with the lag?
Also maybe try running the tool async.
Thanks @wintry jackal, I will try the asynchronous and see if that’s better. I had a similar setup where it would collect all, then populate, but users want to see it real time.