Skip to main content
Skip table of contents

InquireAI Node

This node is available for GlobalCapture and GlobalAction and can be downloaded from the Square 9 Solutions Delivery Network.

InquireAI is only available for Square 9 Cloud solutions.

This node requires additional licensing. There are two types of licensing available:

  • Header Only

  • Header and Line Item

All Square 9 Cloud customer are provided with a one time, 500 page trail of Inquire AI.

Reach out to your Square 9 representative or email us at info@square-9.com for additional assistance or questions.

View Square 9’s Ethical AI policy.

Configure the InquireAI node with a well-structured prompt designed to extract relevant data from multiple variations of the same document type, streamlining the process by removing the need to build and maintain separate templates for each format.

Inquire AI Node

Before You Get Started

You will need to create and add process fields to the Process Field pane with names that match the keys returned by the prompt.

If I use this prompt:

“You are a knowledge worker tasked with understanding payments.  Carefully read the PDF file and extract the following fields: Check Number, Check Date, and Amount.”

I will need Process Fields called Check Number, Check Date, and Amount. The data extracted by the AI will automatically be added to these matching process fields.

Node Properties

InquireAI Settings

Title

Add a title for this node. Your title should be brief but descriptive about what the prompt is doing. Titles are useful when reviewing the history of a workflow for easier understanding of the overall process.

Description

Provide a synopsis of what this node is doing, or make note of any important details. This is useful for providing additional information such as types of documents being ingested, the data being extracted by the prompt, destination archive, etc. A good description is helpful when returning to modify the workflow in the future.

License Key

Enter the Inquire AI license key provided by Square 9.

InquireAI is a InquireAI document extractor. It requires a cloud subscription and a valid license for use. For trial usage, leave the License Key blank.

Log Output

When enabled, AI responses are logged in the workflow process history. A JSON structured output of the data extracted can be viewed by expanding the Inquire AI card in the Process History of the Batch Manager. This is useful for understanding what data is being returned by the AI when data is not appearing in the process fields.

All data returned is returned as a string. Formatting of the returned data may be required to be used in other types of GlobalCapture fields.

More details about the response data can be found on our InquireAI Response Data page.

Clear Table

Enabled - Clears all data in the data table indicated in the Table text field of the settings before InquireAI data is added to the table.

Disabled - Any data extracted by the AI engine is appended to any existing table data.

Process History Log Output

Page

Specify the page to extract data from.  If left blank, all pages will be checked. Use this to minimize the allotment of pages consumed when not all pages need to be checked for data.

 If blank, all pages will be checked and will be counted against page allotment.

Table

To use Multi Value or Table fields, you must be licensed for Line Items.

Multi Value Fields - If extracting data to a multi value field, add the name of the multi value field in the Process Field pane that will accept the data.

Table Fields - If extracting line item data, add the name of the data table in the Process Field pane that will accept the data.

The name of the multi value field or the fields that make up the table should match what is being returned by the AI.

Prompt

Good prompt writing will often require tweaks to the prompt during testing to maximize the success of data extraction.

Enter the prompt you would like to use to extract data. Writing a prompt can be intimidating at first but just like any other skill, can be learned and improved over time. Here are some tips for writing an effective prompt.

  1. Provide Clear Instructions - Be explicit about what the AI should do when inspecting the document. This includes specifying the type of data to be extracted, the desired output format, and conducting a thorough review of the entire document.

    1. Ex. Thinking as an knowledge worker tasked with processing new and used car sales deal jackets, read the entire document and extract the following information: Customer Name, VIN Number, Date of Sale, Make, Model, Vehicle Type, and Purchase Type. Collect all occurrences of the VIN number and compare them to ensure they are identical.

  2. Provide Additional Details - Add additional details about specific fields that need to be extracted that may not be returning the desired results.

    1. Ex. Purchase Type refers to how the vehicle was purchased. This is often listed at Cash, In House Financing, External Financing, or Leased.

  3. Vary the Sequence of Your Phrasing - If the data being returned is not what is expected, it may be necessary to change the ordering of the prompt.

    1. Ex. Thinking as an knowledge worker tasked with processing new and used car sales deal jackets, read the entire document and extract all instances of the VIN Number. Compare the extracted VIN numbers and ensure they are the same. Next, extract the following information: Customer Name, Date of Sale, Make, Model, Vehicle Type, and Purchase Type.

  4. Anticipate Variability - Specify how InquireAI should handle missing or inconsistent data.

    1. Ex. If the VIN Number is not present, return null.

  5. Avoid Ambiguity - If similar terms or sections exist in the document, guide InquireAI on what to ignore or prioritize.

    1. Ex. Only extract the Customer Name from the ‘Primary Address’ section, not from the ‘Delivery Address’.

Full Screen Prompt TextArea ( expand.png )

Click the Full Screen ( expand.png ) icon to the right of the Prompt to expand the prompt textarea to full screen making it easier to see and write your prompt. Press ESC on your keyboard to exit prompt full screen.

Data Validation

Data Validation is only an option in GlobalCapture.

Data Validation

The Data Validation checkbox is enabled by default to enforce Data Types or Length settings for Fields.  When enabled, if data does not meet the type or length settings for the field, the process will error on this node.  

Outputs

The Inquire AI node requires two outputs: Success and License Error.

Success

The document will follow this path if the prompt is successful in communicating with the InquireAI server and data is returned.

License Error

The document will follow this path if there is an issue with your licensing. This will occur if you have exceeded your allotment of pages for the day, you have consumed your one time trial of 500 pages, or you are attempting to do line item extraction without line item extraction licensing.

Best Practices

  • Complete all Inquire AI extraction before any validation, data entry, or template application.

  • If the Key from the Key:Value pair does not match your index field, create a process field that matches the Key and use the Set Process Field node to match the returned Key to the desired process field.

  • When collecting data or asking for a summary that might be long, include a limit to the returned value to ensure it does not exceed the length of the destination process field.

    • Summarize the document in 50 words or less.

    • Limit the Item Description to 150 characters or less.

Use Cases

Classification and Summary

Leveraging InquireAI, document classification becomes a trivial task.  Provide a description of the documents you expect to analyze, ask the AI to summarize the document, and you are ready to go.

Prompt

You are responsible for classifying and summarizing documents being received.

  • Summarize the document 100 words or less and assign it to the Summary field.

  • For classification purposes, documents received may be a Purchase Order (a request for goods and or services), a Contract (a legal agreement outlining terms between two parties), or Meeting Minutes (an outline or summary of a prior meeting). Classify the document type and assign it to the Document Type field.

Using this prompt, you will need the following process fields to access the data in your workflow:

  • Summary

  • Document Type

Purchase Order/Sales Order

Eliminate the need for multiple complex templates to extract header and line item information from Accounts Payable/Receivable documents using Inquire AI. As a general rule, it’s best to be as descriptive as possible, but oftentimes, a very simple prompt will suffice.

Simple Prompt

You are a knowledge worker tasked with understanding purchase orders and sales orders.  Carefully read the PDF file and extract header fields that will aid in management of these document types.  Extract the following fields:

  • Order Number

  • Order Date

  • Ship To

  • Bill To

  • Amount

  • Description

The following are line item fields:

  • Item Description

  • Quantity

  • Unit Price

  • Extended Amount

Descriptive Prompt

You are a knowledge worker tasked with understanding purchase orders and sales orders.  Carefully read the PDF file and extract header fields that will aid in management of these document types.  Extract the following fields:

  • Order Number

  • Order Date

  • Ship To

  • Bill To

  • Amount

  • Description

The following are line item fields:

  • Item Description

  • Quantity

  • Unit Price

  • Extended Amount

Order Number represents a value that normally has a key or order, order number, po number, or sales order number.  Sometimes the value is a derivative of those keys.

Order date represents the date the order was placed.  Orders may sometimes be placed by non-us entities, and in those cases, dates may be presented in non-us formats.  Always assume dates are to be normalized as US date formats unless there is reasonable suspicion the document originated from a non-us entity.

Ship To is the name and address where the goods are to be sent.  All the lines of this value should be returned as a single string.

Bill To is the name of the company who produced the document.  This is often the same company name that is found as the company logo on the document.  This field would only ever be the name, never a complete address.

Amount is the total amount of the order.  If no total amount is found, the amount may be calculated by summing all the amount values of any lines with dollar amounts in a table.

Description should be a summary of the order's purpose in 100 words or less

Extract line items for the following fields:

ItemDescription is a field that might represent the item number, a product description, or both.  If both fields are found, combine them into a single string value.

Quantity is a numeric only value indicating the number of items ordered or purchased.

Unit Price is the dollar amount of a single item

Extended Price is a dollar amount that is generally specifically expressed, but is a calculation of the quantity multiplied by the unit price.

Using either prompt, you will need the following process fields to access the data in your workflow:

Regular Process Fields

  • Order Number

  • Order Date

  • Ship To

  • Bill To

  • Amount

  • Description

A Table Containing Process Fields

  • Item Description

  • Quantity

  • Unit Price

  • Extended Amount

Contracts

Contracts are a complex document format that at times can require contextual understanding to collect accurate data points. Using traditional approaches to extraction, this task is often difficult or impossible.  For documents of this nature, it’s best to speak to the AI engine as you would a new employee being trained to read and understand these documents.

Prompt

You are a knowledge worker tasked with understanding business contracts.  Carefully read the PDF file and extract header fields that will aid in management of the contract.  Extract the following fields:

  • Contract Term

  • Effective Date

  • Contract Originator

  • Contract Recipient

  • Description

Instructions:

Contract Term is a numeric value, generally expressed in months, used to identify the duration of the contract.  Whenever possible, express the value as a numeric number of months.

Effective date is the start date of the contract, which might be directly stated on the document, or might be based on the date the contract was signed.

Contract Originator is the name of the business who authored the contract.

Contract recipient is the 2nd party involved in contract execution.  This could be a business or could be an individual in the case of an agreement with a consultant or sole proprietor.

Description should be a summary of the contracts purpose in 100 words or less

Countersigned indicates whether or not the contract is fully executed by all parties.  Contract signatures are generally found on a single page where there is a signature block for the contract originator and the contract recipient who is often referred to as the customer.  Find the signature page and look for multiple signatures on one page.  When there are unsigned signature lines present, the value should be No for Countersigned indicating the contract has not been executed by all parties.  If all signature lines contain a signature, the value should be Yes.

Using this prompt, you will need the following process fields to access the data in your workflow:

  • Contract Term

  • Effective Date

  • Contract Originator

  • Contract Recipient

  • Description

  • Countersigned

Handwritten Documents

Handwritten Applications, regardless of their use are a problem for any traditional extraction technology, depending on the need, you may run into some challenges, including:

  1. Needing to Extract Handwritten Data (from varying forms/quality of handwriting)

  2. Needing to read Checkboxes

  3. Identify if a document is signed

  4. Normalize data such as dates, regardless of how the data is written.

Prompt

You are a Homestead Tax Exemption Extractor. You will take careful care to read data off of incoming documents. This data will be returned as header data as a JSON. You will extract header values off of documents with great care, below are the data points to look for:

  1. Owner Name 1 - Label/Key as "Owner 1"

  2. Owner Name 2 - Label/Key as "Owner 2"

  3. Property Address -  Label/key as "Property Address"

  4. Mailing Address -  Label/key as "Mailing Address" - If it's indicated it's the same as the property address, use the property address.

  5. OPA Account Number. - Label/key as "Account Number"

  6. Phone Number - Label/key as "Phone Number" - Format as ###-###-####

  7. Document Date, if no Document Date is available, use the Date Signed. (Formatted as MM/dd/yyyy) Label/key as "Document Date")

  8. Email Address - Label/key as "Email".

  9. Is this Property your primary residence?  Read if Yes or No is checked, and return "Yes" or "No" if both are checked, return "Both" if none are checked return "None".  - Label/key as "Primary Residence"

  10. Do you claim anywhere else as your primary residence?  Read if Yes or No is checked, and return "Yes" or "No" if both are checked, return "Both" if none are checked return "None".  - Label/key as "Other Primary Residence"

  11. Is this residence part of a cooperative where some or all the taxes are paid jointly?   Read if Yes or No is checked, and return "Yes" or "No" if both are checked, return "Both" if none are checked return "None".  - Label/key as "Cooperative"

  12. If "Yes" is checked for "Is this residence part of a cooperative where some or all the taxes are paid jointly? " read the percentage and return as a percentage ie: 50%.   - Label as "Cooperative Percentage"

  13. Is part of the property used as a business or rental property?  Read if Yes or No is checked, and return "Yes" or "No" if both are checked, return "Both" if none are checked return "None".  - Label/key as "Rental Property"

  14. If "Yes" is checked for "Is part of the property used as a business or rental property?" read the percentage and return as a percentage ie: 50%.   - Label as "Rental Property Percentage"

  15. Signed - If the document has been signed, indicate a "Yes", otherwise a "No". - Label/key as "Signed"

16. Return the Date the document was signed, you will find that date below the signature. - (Formatted as MM/dd/yyyy) - Label/key as "Date Signed"

Using the prompt, you will need the following process fields to access the data in your workflow:

  • Owner 1

  • Owner 2

  • Property Address

  • Mailing Address

  • Account Number

  • Phone Number

  • Document Date

  • Email

  • Primary Residence

  • Other Primary Residence

  • Cooperative

  • Cooperative Percentage

  • Rental Property

  • Rental Property Percentage

  • Signed

  • Date Signed

Other Uses

Email

Using S9 Notation, use the data collected by InquireAI in an email. This can be useful as a notifier of receipt of a document, a summary of the document, etc.

Email Node Settings

Summary Email

Data Export

Take advantage of data export nodes available in GlobalCapture to export extracted data.

Important Notes

  • Case does not matter. If the prompt asks for Check Number but the index field is check Number, the data will correctly be added to the field.

  • Any data entered into a process field will be overwritten with the data returned by the AI if the process field is included in the response. Fields NOT included in the response will not be impacted.

Process Field

Data Prior to AI

AI Response

Data Post AI

Check Number

1234

“Check Number” = “6789”

6789

Vendor Name

Square 9

“Vendor Name”:””

Invoice Number

123456789

“Invoice No”:”987654321”

123456789

Document Date

05/04/2025

05/04/2025

  • The most common reasons the process will error on the InquireAI are shown below. Be sure to include proper error handling in your workflow.

    • If the AI returns data to a field that is a list and that list does not contain the returned data.

    • If the returned data exceeds the field length. This most common when asking for a summary or extracting an item description.

    • If there is a data type mismatch between the data returned and the field with the same name.

  • All data is returned as type character. If the data returned does not match the type of GlobalCapture field, the process will error. Possible ways to counter this are:

    • Turning off Data Validation will allow the data to populate the field without an error, but the data will need to be corrected before release.

    • Adding details in you prompt about formatting of the data.

Version

Date

Notes

1.3

3/20/2025

Initial feature release

JavaScript errors detected

Please note, these errors can depend on your browser setup.

If this problem persists, please contact our support.