Azure OpenAI
Last updated
Last updated
To have an assistant in the portal, you need to configure the following steps:
Validate the prerequisites in the Power BI Admin Portal (you are here).
Contract the model on Azure OpenAI or OpenAI (ChatGPT)
Create a model on Power Embedded
Create an assistant on Power Embedded
Requirements for Creating an Assistant on Azure,
Have a paid subscription
PTo contract the AI model, it is essential to opt for a paid subscription, as the free version has a very low token limit, making it incompatible with Power Pilot. If you do not have a subscription yet, we can help you. As Microsoft Partners, we offer support to make the subscription available quickly. Simply fill out this form so that our cloud team can contact you: Azure Partnership.
To create the AI access resource, click on this link.
To create the AI access resource on Azure, click on this link.
Now you need to fill in the necessary information for contracting the resource.
Once you move to the next screen, you can leave the first option selected.
In this tab, if you have tags to identify resources based on company configurations, you can associate them. If not, leave them blank
After verifying the data and settings, simply create the resource by clicking on "Create".
When creating the resource, we need to go to the "Keys and Endpoint" tab to copy two important values that we will use when creating the AI model in the Power Embedded admin portal.
With the resource created, we now need to define which model we will use click on this link.
On this new screen, go to Deployments > Create a deployment.
Choosing the AI model is an essential step in configuring your virtual assistant. In this documentation, we opted for the GPT-4o-mini model due to its more affordable cost per token. Click and learn more
You can set the token rate limit according to your assistant usage needs. However, to ensure a smoother and more efficient experience, it is recommended to maintain a minimum limit of 100K tokens per minute.
Done, on the Azure portal, we have completed the entire process; now we will go to the Power Embedded admin portal to create the model and assistant.