Hi OpenAI team,
Currently, Assistants and the Chat Completions API offer the ability to use tools in chat which has greatly improved the use cases for GPT in apps. However, using tool calling in both assistants and completions is rather cumbersome and requires about 3 steps for calling tools.
What would be amazing is if the Assistant’s API could call server endpoints with the created JSON automatically and get the return value (much like how plugins worked). For many developers, this would be so much easier than having to go through the entire tool call process which can be very error-prone at times. It would be great to paste in your endpoint, pass in the arguments and let GPT call the endpoint.
If you guys built this functionality, I would switch to the Assistant’s API full-time and even be willing to pay more for “enhanced tool calling” let’s call it.
Thanks again for all the great work and looking forward to the new 3.5-turbo launch.