AI Extract

The AI Extract node pulls specific information out of text using AI. Give it a message and tell it what to find, and it returns just that piece of data.

How It Works

1. You provide the source text (usually a customer message).
2. You describe what to extract.
3. The AI reads the text and returns only the requested information.

Configuration

Input Text

The source text to extract from. Supports {{variables}}.

Example: {{incoming_message}}

Extraction Prompt

Describe what you want to pull out of the text.

Examples:

• "Extract the customer's email address from this message"
• "Pull out the order number mentioned in this text"
• "What date or time is the customer asking about?"
• "Extract the product name the customer is asking about"

Output

The extracted value is available as {{extracted}} for downstream nodes.

Use Cases

Extract an Email Address

A customer sends: "Sure, you can reach me at [email protected] for the quote."

Set the extraction prompt to: "Extract the email address from this message."

The node outputs [email protected] as {{extracted}}, which you can then save to their contact record using an Update Contact node.

Pull Out an Order Number

A customer sends: "I need help with order #58291, it hasn't arrived yet."

Set the extraction prompt to: "Extract the order number from this message."

The node outputs 58291 as {{extracted}}.

Identify a Date

A customer sends: "Can I book an appointment for next Thursday at 2pm?"

Set the extraction prompt to: "What date and time is the customer requesting?"

The node outputs something like next Thursday at 2pm as {{extracted}}.

Tips

• Be specific about what you want. "Extract the email" is better than "Extract the important info."
• If the text might not contain what you're looking for, you can follow this node with an If/Else or AI Condition to handle cases where nothing was found.
• You can chain multiple AI Extract nodes to pull out different pieces of information from the same message.