What's New

Workflows now include a native AI-powered action for extracting structured data from unstructured text. AI Data Extract takes raw input – emails, SMS messages, webhook payloads, AI outputs, and returns clean, typed variables ready for use in any downstream action.

Define the fields you need, point the action at your input source, and the AI handles the rest. No code & no external tools.

Key capabilities:

• Custom Value Input — Select any text-based data available in your workflow as the input source using the custom value picker.
• Optional Context Field — Provide supplementary context to improve extraction accuracy for complex or ambiguous inputs.
• Pre-Built Templates — Choose from ready-made templates covering common extraction patterns, or define a custom field set from scratch.
• Downstream Ready — Every extracted field becomes a workflow variable, immediately available in subsequent actions.

How to Use It

1. Click "+ Add Action" in your workflow.
2. Search for "AI Data Extract" inside the AI action category.
3. Select your input source via the custom value picker (e.g., email body, webhook payload).
4. Optionally, add supplementary context to guide the extraction.
5. Choose a pre-built template or define your own custom fields.
6. Save. The extracted fields are now available as variables in all downstream actions.

Why This Matters

Extracting usable data from unstructured text has been one of the most consistent customer requests. Until now, doing this inside workflows required manual workarounds or external tools. AI Data Extract closes that gap natively – turning any raw text into structured, actionable variables without leaving the workflow builder.

What's New

Workflows now include a native AI-powered action for extracting structured data from unstructured text. AI Data Extract takes raw input – emails, SMS messages, webhook payloads, AI outputs, and returns clean, typed variables ready for use in any downstream action.

Define the fields you need, point the action at your input source, and the AI handles the rest. No code & no external tools.

Key capabilities:

• Custom Value Input — Select any text-based data available in your workflow as the input source using the custom value picker.
• Optional Context Field — Provide supplementary context to improve extraction accuracy for complex or ambiguous inputs.
• Pre-Built Templates — Choose from ready-made templates covering common extraction patterns, or define a custom field set from scratch.
• Downstream Ready — Every extracted field becomes a workflow variable, immediately available in subsequent actions.

How to Use It

1. Click "+ Add Action" in your workflow.
2. Search for "AI Data Extract" inside the AI action category.
3. Select your input source via the custom value picker (e.g., email body, webhook payload).
4. Optionally, add supplementary context to guide the extraction.
5. Choose a pre-built template or define your own custom fields.
6. Save. The extracted fields are now available as variables in all downstream actions.

Why This Matters

Extracting usable data from unstructured text has been one of the most consistent customer requests. Until now, doing this inside workflows required manual workarounds or external tools. AI Data Extract closes that gap natively – turning any raw text into structured, actionable variables without leaving the workflow builder.

Overview

A live testing capability available on every external-app action step inside the Workflow Builder. Builders can now validate inputs, execute the action as a real call against the connected app, inspect both the exact payload sent and the raw provider response, and have the response schema automatically populate output fields available to subsequent steps.

Test Action closes the gap between configuration time and runtime.

What’s New

• A Test Action button on every external-app action step
• Click Test Action from the action configuration card to launch the test flow. The button is available on every external-app action.

Live Test execution

• The action runs against the connected app as a real API call, using the resolved input values from the builder
• Workflow variables, custom values, and transformations are applied exactly as they would be at runtime
• The connection used is the same Connected Account the action will use when the workflow runs
• Side-by-side Data In and Data Out panes
• Data In — the resolved inputs that will be sent to the provider, after variable interpolation and any transformations.
• Data Out — the raw JSON response from the provider. Large payloads are scrollable and key fields surface as inline chips and cards above the expandable JSON.

How It Works

• Configure the action normally — pick the integration, choose the action (e.g. Find Contact in Klaviyo), and map the inputs.
• If the action uses custom values produced by previous steps in the workflow, the builder requires those upstream values to be resolvable before Test Action can run. The Test button is enabled only when all referenced custom values can be supplied.
• Click Test Action. The Test Drawer slides in from the right with a progress state.
• The live call runs. On success, Data In and Data Out populate the drawer; key fields surface as chips and cards above the expandable JSON.
• Review the response.

What’s new?

You can now connect Browse AI directly to Workflows. One instant trigger — New Completed Task — plus four actions — Run Task, Bulk Run Tasks, Get Task, and Get Bulk Run — let your automations react to Browse AI robot runs and trigger new ones on demand.

How it works

The trigger fires the moment a Browse AI robot task finishes (filterable by Robot), returning the captured data plus task metadata. The four actions let workflows run a single robot, queue a bulk run, or fetch the results of a previously-initiated task or bulk run. Workflow variables are accepted in every input field.

How to use it

• In Workflows, add the Browse AI trigger or action.
• Click Connect your account and paste your Browse AI API key (Settings → API Keys in your Browse AI dashboard).
• For triggers: pick the Robot to monitor and optionally a Select Operator filter, then run Find New Records to capture a sample payload.
• For actions: pick the Robot ID and pass any input parameters the robot expects; run Test Action to validate.
• Save and publish.

Why we built it

Web scraping and content monitoring used to live in their own tab — a Browse AI dashboard separate from the rest of the automation stack. Bringing Browse AI into Workflows lets a customer-submitted URL trigger an on-demand scrape, a daily competitor monitor pipe straight into an internal alert, and a CSV upload of N URLs fire a bulk run with the captured data flowing back into the CRM.

Use Case

A customer submits a competitor URL through a request form. The workflow runs the relevant Browse AI robot against the URL, waits for the task to finish (via the New Completed Task trigger), parses the captured fields, and writes them onto the customer’s CRM record — no manual scraping, no copy-paste.

Additional Notes

• Trigger: New Completed Task (Instant, webhook-backed). Filterable by Robot and Select Operator.
• Actions: Run Task, Bulk Run Tasks, Get Task, Get Bulk Run.
• API key authentication — generate one at Browse AI dashboard → Settings → API Keys.
• Premium trigger and actions — billed per execution at the standard automation rate. Browse AI usage (robot credits / task credits) is billed by Browse AI directly on your account.
• Beta. Multi-account support via the Connected Account dropdown on each step.

What’s new?

You can now connect Browse AI directly to Workflows. One instant trigger — New Completed Task — plus four actions — Run Task, Bulk Run Tasks, Get Task, and Get Bulk Run — let your automations react to Browse AI robot runs and trigger new ones on demand.

How it works

The trigger fires the moment a Browse AI robot task finishes (filterable by Robot), returning the captured data plus task metadata. The four actions let workflows run a single robot, queue a bulk run, or fetch the results of a previously-initiated task or bulk run. Workflow variables are accepted in every input field.

How to use it

• In Workflows, add the Browse AI trigger or action.
• Click Connect your account and paste your Browse AI API key (Settings → API Keys in your Browse AI dashboard).
• For triggers: pick the Robot to monitor and optionally a Select Operator filter, then run Find New Records to capture a sample payload.
• For actions: pick the Robot ID and pass any input parameters the robot expects; run Test Action to validate.
• Save and publish.

Why we built it

Web scraping and content monitoring used to live in their own tab — a Browse AI dashboard separate from the rest of the automation stack. Bringing Browse AI into Workflows lets a customer-submitted URL trigger an on-demand scrape, a daily competitor monitor pipe straight into an internal alert, and a CSV upload of N URLs fire a bulk run with the captured data flowing back into the CRM.

Use Case

A customer submits a competitor URL through a request form. The workflow runs the relevant Browse AI robot against the URL, waits for the task to finish (via the New Completed Task trigger), parses the captured fields, and writes them onto the customer’s CRM record — no manual scraping, no copy-paste.

Additional Notes

• Trigger: New Completed Task (Instant, webhook-backed). Filterable by Robot and Select Operator.
• Actions: Run Task, Bulk Run Tasks, Get Task, Get Bulk Run.
• API key authentication — generate one at Browse AI dashboard → Settings → API Keys.
• Premium trigger and actions — billed per execution at the standard automation rate. Browse AI usage (robot credits / task credits) is billed by Browse AI directly on your account.
• Beta. Multi-account support via the Connected Account dropdown on each step.