How To Use Extractions

Overview

This guide explains how to set up and use custom extractions in Wellgrow software. Custom extractions allow you to capture and store specific user information, such as project details or specific inquiries, tailored to your needs.

1. Setting Up Custom Fields

Navigate to Custom Fields: Go to the Custom Fields section in the software.

Create a Folder: Add a folder to organize your fields. Example: "Client History"

Add Custom Fields:

• Create a new custom field.
• Select Single Line.
• Name the field appropriately.

Example: "Past Experiences"

• Click Save

2. Adding a Custom Extraction Tool

1. Go to AI Assistant: Navigate to Assistants > Tools and Abilities.

2. Create a Tool:

• Select Add a Tool >  Extractions > Create Tool.

Name the Tool: Use a JSON format for naming (all lowercase, with underscores for spaces). Example: past_experiences.

Describe the Tool: Add a clear description explaining what the tool does and when it should be used.

Example:

Use this tool to record the user's past attempts and experiences when it comes to reaching their goals to reference the information at a later date.

3. Choose Custom Field:

• Sync your custom fields with the assistant.
• Select the relevant custom field (e.g., "Past Experiences").

4. Save the Tool:  

• Finalize and save the tool.

5. Example: Creating Another Tool

Tool Name: project_details

Description:

Use this tool to record details about the user’s project needs to reference at a later date.

Custom Field:

 Select or create a field named 

"Project Details".

Add Tool to Assistant: Ensure the new tool is available to the assistant by linking it.

6. Adding Tools to Prompts

1. Format in JSON: Always refer to tool names in JSON format when calling them in prompts. Example: client_issue

2. Example Prompt:

• Client Issue: Based on the user's response, extract the specific roofing service they are inquiring about by calling the function client_issue.
• Project Details: Extract additional details about the project by calling the function project_details.

Dynamic Interaction: The assistant can ask relevant questions and store responses in the designated custom fields.

7. Testing Your Setup

• Ensure all tools are correctly linked to the assistant.
• Simulate conversations to test whether the custom fields are populated with extracted data.
• Verify JSON calls are functioning correctly in prompts.

Conclusion

Custom extractions allow you to gather specific information based on user interactions and store it for future reference. Ensure all tools are named correctly in JSON format, synced with custom fields, and properly integrated into the assistant’s workflows. If you have additional questions, feel free to ask!