Power of GPT in Power Platform

Power of GPT encapsulated in a very simple interface.

?I have been using AI Prompts since few weeks, my team has also built an excellent use case using it, and I have to say I am impressed. It’s a simplest way to use GPT (3.5 and 4o, only two available as of writing this post) capabilities in Power Platform.

You will find AI prompts tab on Power Apps home page (time to contact your Admin to explore this amazing piece of technology)

?

?

It is an interface in Power Platform environment to access GPT capabilities, as you can see in the screenshot there are few templates available to begin with.

You just need to create a Prompt using a very simple interface as shown below, and once done it is ready to use in your Copilot Studio or Power Automate via AI Builder action.

The minimum requirement is an Input, you can define what you want in natural language, whether you want to summarize text or sentiment analysis just type it in under the Prompt section.

Once done click on Test Prompt and check the response under the Prompt response section. If response is not as per your expectation, then work again on your Prompt and repeat this process until you are satisfied with the output.

Prompt setting –

Input: here you can define your Input name and your can test your prompt with sample data. This input name will be used by Power Automate or Copilot Studio when we will consume this.

?

Data used (preview):

It’s a preview setting, but it is wonderful to use, but as of now only data source available is Dataverse.

Output:

Here you can decide whether you want output as Text or as Jason. Jason is very useful if you want more control on the output, for instance to put data in a particular format in Adaptive card it is super useful.

?

Settings:

And this is the brain, as of now only two models are available, GPT 3.5 and GPT 4o. With this you can also set Temperature, in my case I needed predictable results so kept it at 0. ?

?

Overall, I find it very impressive, you can achieve the things in no time which usually may require lot of effort and testing using other ways. Apart from summarization, sentiment analysis and other capabilities, what I found most impressing is that it can be used to generate run time queries like queries for SQL, Odata query for SharePoint etc.

Diwakar Jadhav