Unanswered

Intergating ChatGPT OpenAI with HTTP request in PVA

(0) Share

Report

Posted on by Sonnt

I am setting up the OpenAI ChatGPT configuration using the API through an HTTP request but encountered this error. The parameters have been set up and successfully tested on Postman; however, when trying on PVA (Power Virtual Agents),
I get the error "Sorry, something went wrong. Error code: SystemError. Conversation ID: 6a2e80ed-0616-481c-a90d-f57a0df49cd8. Time (UTC): 11/9/2023 4:08:20 AM." I will provide the HTTP configuration image below.

Categories:

Calling actions from Copilot Studio

I have the same question (0)

All responses (8)

Answers (1)

DamianSu 10 on at

Like (1)

Report

Hello, I have the same problem. I'm trying to integrate Language Studio Custom Question Answering but it's not working. From Postman and even in Bot Framework Composer it works normally. I tried every possible option to do it...

Was this reply helpful? Yes No
Sonnt on at

Like (0)

Report

Dear Damian,
We are waiting for this issue from Microsoft experts.
Thank you

Was this reply helpful? Yes No
HenryJammes on at

Like (0)

Report

Hi @Sonnt, @Damian2,

Not 100% sure it's the same issue, but there is one known issue for which a fix is coming.
The error occurs when you use JSON in your request's body.

As a workaround, instead of JSON, can you convert the content to Formula (Power Fx) and try again?

Henry

Was this reply helpful? Yes No
Sonnt on at

Like (0)

Report

Dear @HenryJammes ,
I already converted the content to Formula (Power Fx) as your suggestion, but the same this issue.
If you tested for succession, please to share for us,
Thank you so much!
Best,
Sonnt

Was this reply helpful? Yes No
HenryJammes on at

Like (0)

Report

Question: can you check how you wrote the "Content-Type" key in your header?
API Reference - OpenAI API

Was this reply helpful? Yes No
Sonnt on at

Like (0)

Report

Hi @HenryJammes
Thank for you suggestion,
I changed the "Content-Type" key in my header, it is the same original OpenAI header (figure 1), but the same issue (figure 2-3).
I have an ideal about my issues, can you suggest me about 'Save response as' in HTTP request based on Open AI responses.
{
"id": "chatcmpl-123",
"object": "chat.completion",
"created": 1677652288,
"model": "gpt-3.5-turbo-0613",
"system_fingerprint": "fp_44709d6fcb",
"choices": [{
"index": 0,
"message": {
"role": "assistant",
"content": "\n\nHello there, how may I assist you today?",
},
"finish_reason": "stop"
}],
"usage": {
"prompt_tokens": 9,
"completion_tokens": 12,
"total_tokens": 21
}
}
Thanks you so much!
Best,
Sonnt,

Was this reply helpful? Yes No
Verified answer

Sonnt on at

Like (1)

Report

Dear @Damian2, @HenryJammes
You should be change response data type as "Get from sample data"
and Go to Test and select a value variable.
Thank you!
Best,
Sonnt

Was this reply helpful? Yes No
jpiabrantes 29 on at

Like (0)

Report

I am also trying to call OpenAI API from CoPilot Studio. Is there a way to use the user message in Body (JSON) of the HTTP Request?

Was this reply helpful? Yes No