Skip to main content

AI Action Steps in Automations

Use AI in your Streak automations to automatically trigger data entry, deal summaries, deal queries, and more.

Brita Ulf avatar
Written by Brita Ulf
Updated today

Streak's automations let you add AI-powered actions that run automatically when a trigger fires.

There are four ways to use AI action steps in your automations: AI Autofill box fields, Ask AI about a box, Generate a box summary, or write a prompt to run a custom AI action.

Each consumes AI credits when the automation runs.

How to add an AI action step

  1. Open Integrations & Automations from the toolbar in Streak Home view.

  2. Select Create native integrations from scratch to choose your trigger and action steps.

  3. Name your automation by clicking in the text field at the top of the automation panel.

  4. Click + Add trigger and select the trigger type from the menu.

  5. Click + Add step in the lower right corner.

  6. Select the Streak option and choose your AI action step.

  7. Configure the action's fields (e.g., select fields to autofill, write your question, etc.).

  8. Use the preview to test with sample data — previews are free and do not consume credits.

  9. Click Create to save your automation.

AI action steps

There are four AI action steps available in Streak automations.

1. AI Autofill box fields - Streak action

Automatically runs AI autofill on selected fields in a box. Go to the pipeline view to write custom autofill prompt instructions for each field.

Configuration:

  • Field names (required): Select which fields to autofill. A suggestions dropdown shows all available fields in your pipeline. You can select as many as needed.

  • Skip filled cells: When enabled, autofill won't overwrite fields that already have data — it only fills empty cells.

Example use case: When a new box is created, automatically populate company info, deal size, and industry fields based on email thread context.

2. Ask AI about a box - Streak action

Asks AI a question about a box and returns an answer each time the automation runs.

Configuration:

  • Question (required): Type the question you want AI to answer about each box. This same question runs every time the trigger fires.

  • Answer (output): The AI-generated response. This value can be used by subsequent steps in the automation.

  • Publish answer step: Add a step to publish the answer somewhere. For example, you could post the answer as a comment in the box timeline (seen above), in a Slack message, or in the notes column of your pipeline.

Example use case: When a new email arrives, ask "What are the customer's main pain points?" to automatically extract key issues from the thread and post it as a comment in your timeline.

3. Generate a box summary - Streak action

This AI action generates an AI summary and saves it in your box view and the summary magic column in the pipeline view.

Configuration:

  • Short summary (output): A brief AI-generated overview of the box.

  • Long summary (output): A more detailed AI-generated summary with additional context.

Example use case: When a stage changes, regenerate the box summary so managers always see an up-to-date snapshot of each deal.

4. AI Response - Utility action

Unlike the other three AI action steps — which are specifically designed around Streak boxes and automatically use the full context of the box (emails, fields, notes, etc.) — the AI Response step is generic input/output with an LLM. It doesn't pull in box context on its own; instead, you define exactly what input goes in and what response you want back.

Because the AI Response step doesn't depend on a box for context, it allows you to perform AI automations on emails before they’ve been added to Streak. This helps you process and structure raw email content so downstream steps have clean, usable data.

Where to find it: Add step → Utilities → AI Response (note: the other three AI steps appear under the Streak menu)

Configuration:

  • Input data (required): The data you want the AI to process. This can be mapped from a previous step's output — for example, the body of an incoming email.

  • Prompt (required): Your instructions telling the AI what to do with the input data — for example, "Identify the lead's first and last name from this form submission."

  • Completion (output): The AI-generated response.

Example use case: Parsing an incoming email that you receive every time a lead form is submitted. You could set up:

  1. Trigger: Gmail incoming message

  2. AI Response step: "From the following email, extract the sender's company name.” You can add multiple AI response steps to extract additional information from the same email.

  3. Next steps: Use the AI Response output to populate box fields, add a comment, or route the email to the right pipeline.

Credits: Like the other AI action steps, AI Response consumes AI credits each time the automation runs.

Using AI credits for automations

AI action steps consume credits from your team's shared credit pool each time the automation runs.

  • Previews are free. Credits are only charged when the automation is live and triggered.

  • When credits run out, the AI action step will fail at runtime. Admins will be prompted to purchase more credits or upgrade the plan. Non-admins will see an email sent to their admin.

  • Admins can view credit usage and remaining balance from the billing settings page.

Tips

  • You can chain multiple AI actions in one automation (e.g., "Ask AI about a box" followed by "AI Autofill box fields").

  • Use "Skip filled cells" on Autofill to avoid overwriting data that's already been entered manually.

  • Use the preview feature to test your setup — it won't use any credits.

  • The "Ask AI about a box" answer output can feed into subsequent steps in your automation, like posting a comment, updating a box field, or sending a message in Slack.

Did this answer your question?