How to use Chat-GPT in Power Automate Desktop? Full Tutorial

⚠️⚠️⚠️ OpenAI changed their API - this method still works but you have to make some changes - please check my pinned comment ⚠️⚠️⚠️
Follow this tutorial if you want to learn how to connect to Chat-GPT with Power Automate Desktop. We fetch the results via the API and store them in an Excel file.
⭐️ Links to the resources ⭐
🔗 Click here if you don't have an OpenAI account yet: • How to get an Open AI ...
🔗 Configure the API key in OpenAI: platform.openai.com/account/a...
🔗 Documentation of the API: platform.openai.com/docs/guid...
⭐️ This is the request body of the Invoke Web Service I use⭐
{
"model": "text-davinci-003",
"prompt": "Please provide 10 ideas for an RPA RUclips channel",
"max_tokens": 1000,
"temperature": 0,
"n": 1,
"stream": false,
"logprobs": null,
"stop": null
}
⭐️ Follow me ⭐
💼 LinkedIn: / thomas--janssen
🧡 Instagram: TomsTechAcademy
#powerautomate #powerautomatedesktop #roboticprocessautomation

[@TomsTechAcademy]

Hi, OpenAI changed their API, this method still works - but you have to change some parameters:
URL: api.openai.com/v1/chat/completions
Request Body:
{
"model": "gpt-3.5-turbo",
"messages": [{"role": "user", "content": "Please provide 10 ideas for an RPA RUclips channel"}],
"temperature": 0.7
}
Click on "Advanced" and disable "encode request body"
Please let me know whether this works for you :)

[@renealvarezm]

This is amazing. Thanks for sharing. Can't wait to try myself.

[@TomsTechAcademy]

Thank you for your comment Rene! Happy coding and let me know in case you need any help 🚀

[@AnilAnil-gj1vk]

can you do one video on how to extract the date and time from the Outlook file and create that name and date with a file name

[@TomsTechAcademy]

Sure I would be happy to! Just to double check: you want to extract the date an email was received in Outlook Desktop?

[@FMA-jn9fq]

Amazing !
When I try, I have a problem with "choices" parameter in the display message (my code: %JsonAsCustomObject['choices'][0].text%).
Error Details: The "JsonAsCustomObject" variable doesn't have a "Choices" property.
Do you have any suggestion to solv this problem?

[@0805pati]

Same here

[@TomsTechAcademy]

Hi, OpenAI changed their API, this method still works - but you have to change some parameters:
URL: api.openai.com/v1/chat/completions
Request Body:
{
"model": "gpt-3.5-turbo",
"messages": [{"role": "user", "content": "Please provide 10 ideas for an RPA RUclips channel"}],
"temperature": 0.7
}
Click on "Advanced" and disable "encode request body"
Please let me know whether this works for you :)

[@FMA-jn9fq]

Yes, thanks@@TomsTechAcademy / Regarding "Display Message" action, please put -> %JsonAsCustomObject['choices'][0]['message']['content']% VS %JsonAsCustomObject['choices'][0].text%

[@juanvillagra8212]

Hello Tomas I tried to use "model "gpt-3.5-turbo" but I get error .... "You exceeded your current quota, please check your plan and billing details."
Any free model ??

[@TomsTechAcademy]

When you create a new account you get some credit which you can use for 3 months. It seems this period has expired. You can just top up your account with $5,- and you're good to go :)

[@vitomillauer9756]

Does this also work for free? I get the error message that I have insufficant quota

[@TomsTechAcademy]

Hi, when you sign up for a new OpenAI account, you receive free tokens which are valid for 3 months. So when you create a new account you can use GPT 3.5 for free for the first 3 months. After the first 3 months you have to top up your balance (you can just put $5,- on your balance).
If you want to use newer models like GPT 4 - you also need a premium subscription which will cost you $20,- per month.