ChatGPT (OpenAI)

OpenAI is a artificial intelligence platform that allows developers to integrate advanced language models into their applications. It supports diverse tasks such as text completion, summarization, and sentiment analysis, enhancing software functionality and user experience.

Below you can find the Aembit configuration required to work with the OpenAI service as a Server Workload using the OpenAI API.

Prerequisites

Before proceeding with the configuration, ensure you have an OpenAI account and API key. If you have not already generated a key, follow the instructions below. For more details on API key authentication, refer to the official OpenAI API documentation.

Create Project API Key

1. Sign in to your OpenAI account.

2. Navigate to the API Keys page from the left menu.

3. Click on Create new secret key button in the middle of the page.

4. A pop-up window will appear. Choose the owner and project (if not multiple, the default project is selected). Then, fill in either the optional name field or service account ID, depending on the owner selection.

   • If the owner is selected as You, under the Permissions section, select the permissions (scopes) for your application according to your needs.
   • Click on Create secret key to proceed.

5. Click Copy and securely store the key for later use in the configuration on the tenant.

Note

The following configuration steps will also work with user API keys; however, project API keys are recommended as they offer more granular control over the resources.

Server Workload Configuration

1. Create a new Server Workload.

• Name - Choose a user-friendly name.

2. Configure the service endpoint:

• Host - api.openai.com
• Application Protocol - HTTP
• Port - 443 with TLS
• Forward to Port - 443 with TLS
• Authentication method - HTTP Authentication
• Authentication scheme - Bearer

Credential Provider Configuration

1. Create a new Credential Provider.

• Name - Choose a user-friendly name.
• Credential Type - API Key
• API Key - Paste the key copied from OpenAI Platform.

Client Workload Configuration

Aembit now handles the credentials required to access the Server Workload, eliminating the need for you to manage them directly. You can safely remove any previously used credentials from the Client Workload.

If you access the Server Workload through an SDK or library, it is possible that the SDK/library may still require credentials to be present for initialization purposes. In this scenario, you can provide placeholder credentials. Aembit will overwrite these placeholder credentials with the appropriate ones during the access process.

Access Policy

• Create an Access Policy for a Client Workload to access the OpenAI Server Workload. Assign the newly created Credential Provider to this Access Policy.

Required Features

• You will need to configure the TLS Decrypt feature to work with the OpenAI API Server Workload.