Generative AI

The Generative AI application allows you to query both OpenAI and Azure OpenAI models using data from your Keboola project.

Create New Configuration

Set up a new configuration for the OpenAI application.

Retrieve your API key from the Azure Portal. Follow the steps provided in the Azure OpenAI quickstart guide for guidance.

For Azure OpenAI, obtain your API key from the OpenAI platform settings.
Enter your API key into the API Key field within the Data Source Configuration window.
Specify the API Base URL, which should resemble: https://YOUR_RESOURCE_NAME.openai.azure.com/
Input the Deployment ID of your model, which can be located at: Azure Portal -> Azure Open-AI -> Model Deployments.
Specify the API version you intend to use. For more details, refer to the Azure OpenAI documentation.

Upon completeing these steps, click the Add Row button and provide a name for the row configuration (this is solely for identification).

Click the List Models button to view the available models. Subsequently, choose your desired model in the Model field.
You can either stick with the default model settings or customize them in the Model Options section.
Consider the Prompt Template Sync Action as a template when crafting your prompts.
The Prompt field determines the prompt and data input format. Use the placeholder [[INPUT_COLUMN]] to reference an input column. Ensure your input table includes the mentioned column.
In the Destination section, select a name for your Storage table. Optionally, you can configure incremental load settings. If primary keys are designated, rows in the destination table will be updated. Without primary keys, new rows will be added.