Enter a prompt. For example, this could be one or more questions about the document.

Enter the text you want used to answer the questions.

The message that needs to be routed to one of the routes below.

Specify the name of the route (eg. a person/group/team) in the first column. And a brief description of the route in the second field. Add as many routes as you'd like. Example: Support tier 1: Handles first response of new tickets.

Any free form blob of text with lead data like name, email, phone number, etc. to be extracted from the content. For example, "Thanks for the reply Robert Example, we'll give you a call back ASAP at 214-555-9854!".

Enter a fully qualified web page URL that is publicly available on the internet.

Enter a prompt. For example, this could be one or more questions about the document.

This is the text you want to parse in the action. You'll want to insert data from previous steps of the Zap here (for example, the body of an email, ticket, tweet, etc).

This is the text for the first example of training. EG: "Thank you for your order, your identifier is 589427623, please come again!"

Tip: While we require only 1 example, be sure to provide multiple examples to increase the AI's sense of what you want parsed. Also, experiment with different examples. For instance, if you see a failure in production you can swap out an example with that failure.

This is the expected data for the first example. EG: order_number as the key and 589427623 as the value. Enter as many keys and values as you like.

Tip: More descriptive keys are generally better at clueing the AI into what you want!

This is the text for the second example of training. EG: "We appreciate your business. View your order at https://yourshop.com/orders/97465234. Thank you."

Tip: be sure to provide a real world examples of freeform text that mimic what you'll pass into the "Freeform Text (Input)". This helps the AI pick up patterns.

This is the expected data for the second example. EG: order_number as the key and 97465234 as the value.

This is the text for the third example of training. EG: "Your order (238715146) is being processed! Please call us at +15551256792 if you need help!"

Tip: Be sure to provide a "sneaky" example, like include another number that looks like an order ID but may be a tracking number, or phone number instead. This can teach the AI to be more discerning.

This is the expected data for the third example. EG: order_number as the key and 238715146 as the value.

This is the text for the fourth example of training. Much like the first few, this is optional, and most users do not need this many examples.

Tip: As your Zap runs, you might take failed cases and use them here with the correct usage to make the AI better.

This is the expected data for the fourth example.

This is the text for the fifth example of training. Only use this if you have a fairly complex use case. Most folks only need 2 or 3! Or, try more descriptive keys in your data.

This is the expected data for the fifth example.