Provisioning and setting up the generative AI provider for your application

Before you begin

• Make sure you have the HelixGPT Administrator role to provision your AI provider.
• Specify the API endpoint URL and API key of the generative AI provider that you want to use.
• We recommend setting the Tokens Per Minute (TPM) in your third-party GenAI provider console to at least 100,000 TPM. You can review and adjust this setting based on your usage patterns.
• (For version 23.3.02 and version 23.3.03) Set the Content Filter for the model to Default.

Process for setting up BMC HelixGPT

The following image shows the process of setting up BMC HelixGPT and the current step that you are on:

To add the API endpoint URL and the API key to the generative AI provider

For every Microsoft Azure OpenAI model that you use, you must specify the deployment name, API endpoint URL, and API key separately.

1. Log in to BMC Helix Innovation Studio.
2. On the Workspace tab, from the table, click HelixGPT Manager. 
3. On the HelixGPT Manager page, click Visit deployed application.
4. On the HelixGPT Home page, click Settings .
5. Select HelixGPT > Connections > AI service providers.
   The details of the AI service provider models that are supported by default are displayed.
6. For the Azure OpenAI model that you plan to use, perform the following steps:
   1. Specify the Deployment name, API endpoint URL, API key, and API Version.
      The following image shows an example of the details for the Azure OpenAI service provider:
      

   2. To test the AI service provider settings, click Test Connection.

      Important

      If the connection fails, verify that you have entered the correct Deployment name, API endpoint URL, API key, and API Version.

   3. (Optional) Repeat these steps for every Azure OpenAI model that you plan to use.
7. For the Google Vertex models that you plan to use, perform the following steps:
   1. Specify the Deployment name, Endpoint ID, Credentials, and Location.
      The following image shows an example of the details for the Google Vertex service provider:
      

   2. To test the AI service provider settings, click Test Connection .

      Important

      If the connection fails, verify that you have entered the correct Deployment name, Endpoint ID, Credentials, and Location.

8. Click Save.

To change the authentication type for a model

When configuring a model in BMC HelixGPT, you can change the authentication type from the default API Key to Oauth2.

The following Oauth2 grant types are supported:

• Resource Owner Password Credentials 
• Client Credentials Grant

To learn more about these grant types, see Oauth2 Grant Types.

1. Log in to BMC Helix Innovation Studio.
2. Select Workspace > HelixGPT Manager.
3. Select the Model record definition, and click Edit data.
4. On the Data editor (Model) page, click New to create a new record or select an existing record, and click Edit.
5. In the New record or Edit record pane, from the Auth Type list, select OAuth2.
   Important: The default authentication type is API Key.
6. In the Auth Grant Type field, enter one of the following options:
   • password—if your authentication grant type is Resource Owner Password Credentials Grant.
   • client_credentials—if your authentication grant type is Client Credentials Grant.
7. Specify the Auth Client ID, Auth Secret, Auth Scope, and Auth Url.
8. Click Save.

The following image shows an example of the Azure OpenAI service provider with the authentication type set to Oauth2:

Where to go from here