Connect OpenAI to analyze text and images

Use our no-code step to analyze, generate emails, forms, or chat replies, and process images.

Before you begin

• In your Next Matter portal, go to Automations library, click Connect in the OpenAI app tile, and enter the OpenAI API token to connect.

Use the integration

Analyze or generate text

1. Click Workflows > + Create new… > Create a new workflow and give your workflow a name. If you have a workflow ready, click … > Edit workflow.

2. Click + Add step and select Integrations > OpenAI.

3. Click the step and click Configure integration.

4. Select Process/generate text with AI in OpenAI operation.

5. Enter the prompt and select the AI model and the output type.

   When you choose a reasoning model from the o1 series, it will bypass your Advanced AI configuration settings.

6. (Optional) If needed, you can tweak the request by providing data in the Advanced AI configuration dropdown.
7. Select the output text as a variable you can refer to later on.
8. Save your changes.

Use cases

• You can use the no-code step to, for example, analyze customer emails for the support team, surface the email’s main points, and measure the sentiment.
• You can analyze customer emails to automatically input data into insurance claim forms that customers can review and confirm.

Process images

1. Click Workflows > + Create new… > Create a new workflow and give your workflow a name. If you have a workflow ready, click … > Edit workflow.
2. Click + Add step and select Integrations > OpenAI.
3. Click the step and click Configure integration.
4. Select Process images with AI in OpenAI action.

5. Enter the prompt and select the AI model and the output type.
6. Select the data reference to the images uploaded in any of the previous steps. Currently, we don’t support files in PDF.
7. (Optional) If needed, you can tweak the request by providing data in the Advanced AI configuration drop-down. Tip: You can use floats to define Temperature but make sure to use ”,” (comma) instead of ”.” (point) for the German/ European lokale.
8. Select the output text as a variable you can refer to later on.
9. Save your changes.

Make sure the images do not exceed 30 megapixels each. Large images might cause the step to fail.