Set your AI model for the Autonomous Debugger¶

The Lightrun Autonomous Debugger uses an AI model that analyzes data, recognizes patterns, and makes predictions or decisions to automate and enhance tasks. By default, it is set to use Lightrun’s OpenAI account. This guide outlines how administrators can configure a custom AI model, such as a company OpenAI key or a Private Azure OpenAI deployment in the Lightrun Management Portal.

Supported AI model providers¶

• OpenAI: By default, the Autonomous Debugger uses Lightrun's OpenAI as the default AI model provider. No configuration is required.

• Customer OpenAI: Requires adding your API key.

• Private Azure OpenAI Service: Requires copying a few values from your Private Azure deployment.

Enable the Autonomous Debugger in your organization¶

1. Log in to the Lightrun Administration Portal.

2. In Settings, navigate to the AI Debugger > Global Settings.

   

3. Toggle the AI Debugger switch to enable the feature.

   

   In the AI provider model list, the Lightrun OpenAI account is selected as the default.

4. Click Save

Set your organization's OpenAI account as the AI model provider¶

To use your company’s OpenAI account, you will need to obtain your company’s API Key.

1. Log in to the Lightrun Administration Portal.
2. In Settings, navigate to the AI Debugger General Setting tab.
3. Toggle the AI Debugger switch to enable the feature.

4. From the AI model provider list, select Customer OpenAI account.

   

5. (Optional) In the Custom URL field, enter your URL.

6. (Mandatory) In the API Key field, paste your company’s API Key.
7. Click Test connection to validate that you can connect to your OpenAI account.
8. Click Save.

Set Microsoft Azure OpenAI as your AI model provider¶

You can choose to set up your own OpenAI account using your Private Azure OpenAI service. To set up the model, you are required to enter a number of settings that are required to be obtained in Azure.

Prerequisites¶

• Deploy gpt-4o and text-embedding-3-small modelstabs=global-standard%2Cstandard-chat-completions#embeddings) on your Azure OpenAI service.
• Make sure the new deployments have full capacity allocated. For more information, see Manage Azure OpenAI Service quota.

Stage 1: [Azure AI Foundry] Obtain the Azure API parameters¶

1. Log in to https://ai.azure.com/.

2. In the left navigation bar, click Models + endpoints.

   The Manage deployments of your models and services opens.

3. Click Get endpoint.

   

4. Copy the key.

   

5. Copy the Completion and Embedding deployment names.

   a. In the left navigation bar, click the Models+endpoints page.

   A list of available deployed models is displayed.

   

   b. Copy the gpt-4o model deployment name to be pasted in the Completion deployment name field.

   c. Copy the text-embedding-3-model deployment name, for the Embedding deployment name field. Note that under certain circumstances, the names could be gpt-4o and text-embedding-3-small respectively.

6. Copy the Endpoint. Do not copy the endpoint from the Get endpoint dialog. Instead, follow these steps:

   1. Go to the Azure portal in https://portal.azure.com/.

   2. Click the Azure AI services and then the Azure OpenAI links.

   3. Select the same deployment selected in the previous steps.

   4. In the left navigation pane, navigate to Resource management, and select Keys and Endpoint.

   The Keys and Endpoint dialog opens.

   

   5. Copy the Endpoint which should match the following syntax: https://azure-cognitive-XXXXX.openai.azure.com/.

Stage 2: Configure Azure OpenAI in the Lightrun Management Portal¶

1. Log in to the Lightrun Administration Portal.
2. In Settings, navigate to the AI Debugger General Setting tab.
3. Toggle the AI Debugger switch to enable the feature.

4. From the AI model provider list, select Private Azure OpenAI service.

   

5. Paste the parameters you copied in Stage 1 from your Azure AI Foundry environment.

   Parameter	Description
   Azure API Key	Your Azure Key.
   Azure API endpoint URL	The URL for the Azure Endpoint, which is used for sending API requests.
   Completion deployment name	The deployment name for the Azure gpt-4o model, which is used for generating text completions.
   Embeddings deployment name	The deployment name for Azure text-embedding-3-model, which is used for generating vector embeddings.

6. Click Save.