Skip to main content

Add Prompt Components(JSON Prompt)

Overview​

Accelerate field updates. Reduce manual effort. Improve consistency.

Updating fields manually slows your process and invites errors. GPTfy allows you to automatically populate multiple fields based on AI-generated responses, saving time and improving record quality.

Let’s dive in!

  • Create custom actions for your prompts
  • Fine-tune for length, tone, and language
  • Auto-update fields with AI response data

1. Set Up a JSON Prompt​

  • Navigate to GPTfy → Prompts Tab
  • Click New Prompt → Select Type: JSON
  • Save your new prompt

alt text

2. Add Prompt Components​

  • In the Prompt Components panel, click New
  • A pop-up window appears. Fill in:
    • Component Name
    • Target Object (must match your Mapping)
    • Action (defines what AI should generate)
    • Length, Language, Tone, Format

alt text

Note: Toggle Prefix/Postfix in AI Settings → Preferences.
When enabled, each AI response is automatically wrapped with your configured prefix and suffix text.

3. Configure Prompt Component Advanced Settings​

  • Constraint Value: Define strict word limits, sentiment constraints, etc.
  • Additional Instructions: Give AI extra details.

alt text

Tip: Additional Instructions can now be dynamically entered by users if 'Allow User Input' is enabled, enhancing response relevance.

4. Create and Map Actions​

To auto-update Salesforce fields based on AI outputs:

  • Navigate to the Actions tab on your prompt
  • Click New Action → Choose:
    • Update Field
  • Fill in:
    • Target Object (must match your Prompt Component Object)
    • Field to update
    • AI Response Node (choose which part of AI output to map)
    • Append Timestamp (optional)

⚡ Important: Object in Object Manager → Picklist Values and the Mapping must match to enable correct Action execution.

Note: Users can also update and delete the created prompt component.

5. Test Your Setup​

  • Use the Test tab inside Prompt Builder
  • Select a test record
  • Click Run to validate AI-generated responses and field updates

Additional Best Practices​

  • Always test prompt updates on sandbox records first
  • Use Additional Instructions to fine-tune the output for different contexts
  • Ensure proper Field Mapping in Data Context Mapping to avoid missing data during updates

Key Benefits​

  • Save hours of manual updating effort
  • Ensure consistency in customer-facing communications
  • Empower your teams to move faster with accurate CRM updates