2702 Ai In Forms
AI in Forms
Introduction to Form-Based AI
Types of Form AI
- Single-Field AI - Generate content for one field based on other fields
- Multi-Field AI - Use multiple fields as input to generate content
- AI Prompt Buttons - User-triggered AI generation on demand
Single-Field AI
What is Single-Field AI?
Use Cases
- Product Descriptions - Generate from name, category, features
- Email Subjects - Create from email body content
- Summaries - Condense long text into brief overview
- Titles - Generate from main content
- Tags - Auto-generate tags from content
- Categories - Suggest category based on description
Setting Up Single-Field AI
Step 1: Add AI Field to Table
- Go to Data Builder
- Select your table
- Click "Add Field"
- Choose field type (usually Long Text for AI-generated content)
- Name the field (e.g., "AI Generated Description")
Step 2: Add Field to Form Component
- Go to Page Builder
- Edit the form component
- Add your AI field to the form
- Click the field settings
Step 3: Enable AI
- In field settings, find "AI Prompts" section
- Toggle "Enable AI" on
- Choose trigger: "Auto Generate" or "Manual"
Step 4: Write Your Prompt
- In the AI Prompt field, write your instruction
- Reference other fields using field variables
- Be specific about desired output
Example: Product Description Generator
Table Structure:
- Product Name - Text field
- Category - Connection to Categories table
- Key Features - Long text field
- Price - Number field
- AI Description - Long text field (AI-enabled)
AI Prompt:
Write a compelling e-commerce product description for {Product Name}.
Product Category: {Category}
Key Features: {Key Features}
Price: ${Price}
The description should be 2-3 paragraphs, highlight benefits, and include a call to action. Use an enthusiastic but professional tone.
Result:
When a user creates a new product and fills in the name, category, features, and price, the AI automatically generates a polished product description ready to use.
Auto-Generate vs Manual
Auto-Generate
- Triggers: When record is created or updated
- Best For: Content that should always be generated
- User Experience: Seamless, automatic
- Use When: Output is always needed and input data is reliable
Manual
- Triggers: User clicks "Generate" button
- Best For: Optional content or when user wants control
- User Experience: User decides when to generate
- Use When: Users may want to regenerate or skip AI
Best Practices for Single-Field AI
- Be Specific - Clearly define what you want generated
- Provide Context - Include all relevant field data
- Set Expectations - Specify length, tone, format
- Use Examples - Show the AI what good output looks like
- Field Validation - Ensure input fields are filled before AI runs
- Allow Editing - Let users modify AI output if needed
Multi-Field AI
What is Multi-Field AI?
Use Cases
- Content Expansion: Take brief notes and generate multiple formatted outputs
- Data Enrichment: Extract different aspects of text into separate fields
- Multi-Format Content: Generate long and short versions simultaneously
- Structured Extraction: Parse unstructured text into organized fields
Setting Up Multi-Field AI
Step 1: Plan Your Fields
Identify which fields will be inputs and which will be AI-generated outputs.
Step 2: Create AI Action
- In Page Builder, edit your form component
- Add an Action Link or button
- Choose "AI Prompt" action type
- Configure to update multiple fields
Step 3: Write Prompt with Multiple Outputs
Structure your prompt to generate content for each target field.
Example: Meeting Notes Processor
Table Structure:
- Raw Notes - Long text field (input)
- Summary - Long text field (AI output)
- Key Decisions - Long text field (AI output)
- Action Items - Long text field (AI output)
- Next Meeting Topics - Long text field (AI output)
AI Prompt:
Analyze these meeting notes and extract the following:
Meeting Notes: {Raw Notes}
Please provide:
1. SUMMARY (2-3 sentences of key discussion points)
2. KEY DECISIONS (bullet list of decisions made)
3. ACTION ITEMS (bullet list with person responsible and deadline if mentioned)
4. NEXT MEETING TOPICS (topics to discuss next time)
Format each section clearly with the section headers above.
Configuration:
Set up the action to parse AI output and populate each field with its respective section.
AI Prompt Buttons
What are AI Prompt Buttons?
Use Cases
- Content Variations: Generate multiple versions to choose from
- Enhancement: Improve existing user-written content
- Optional AI: Let users decide if they want AI help
- Multi-Step Process: Generate different content at different stages
Setting Up AI Buttons
Step 1: Add Action Link
- Edit your form component
- Click "Add Action Link"
- Name it descriptively (e.g., "Generate Description")
- Choose icon and styling
Step 2: Configure AI Action
- Select "AI Prompt" as action type
- Write your prompt
- Select target field(s) to populate
- Configure success message
Step 3: Add Multiple Buttons (Optional)
Create multiple buttons for different AI operations on the same form.
Example: Email Draft Generator
Form Fields:
- Customer Name - Text
- Issue Type - Dropdown
- Customer Message - Long text
- Response Draft - Long text
AI Buttons:
Write a professional customer service response to the following:
Customer: {Customer Name}
Issue Type: {Issue Type}
Their Message: {Customer Message}
Create a helpful, empathetic response that addresses their concern. Keep it concise and actionable.
Rewrite this response in a more formal, professional tone:
{Response Draft}
Rewrite this response in a warmer, more conversational tone:
{Response Draft}
Advanced Form AI Techniques
Conditional AI Generation
Use Cases:
- Generate content only for specific record types
- AI runs only when required fields are filled
- Different prompts based on category or status
Implementation:
Use conditional rules or display rules to show/hide AI buttons based on field values.
Chained AI Operations
Example Workflow:
- First AI: Generate draft content from inputs
- Second AI: Enhance and format the draft
- Third AI: Generate summary of final content
Implementation:
Use record rules that trigger subsequent AI operations based on field updates.
AI with Connected Records
Example:
Generate a personalized proposal for {Customer.Company Name}.
Industry: {Customer.Industry}
Previous Orders: {Customer.Order History}
Specific Needs: {Requirements}
Create a proposal highlighting relevant products and pricing.
User-Driven Prompts
Form Setup:
- Add "AI Instructions" text field
- User enters what they want AI to do
- AI prompt includes user instructions
AI Prompt Template:
User Request: {AI Instructions}
Using this data:
{Field1}, {Field2}, {Field3}
Complete the user's request.
Form AI Best Practices
User Experience
- Clear Labeling: Label AI fields clearly so users know what's generated
- Loading Indicators: Show users AI is working (built-in spinner)
- Edit Capability: Always allow users to edit AI output
- Regenerate Option: Let users try again if not satisfied
- Preview: Show AI output before saving when possible
Prompt Design
- Be Specific About Length: "Write 2-3 paragraphs" or "Limit to 100 words"
- Define Tone: "Professional", "Friendly", "Technical", "Simple"
- Specify Format: "Bullet points", "Numbered list", "Paragraph form"
- Include Examples: Show AI what good output looks like
- Use Field Names: Reference fields with {Field Name} syntax
Performance
- Limit Auto-Generation: Don't auto-generate on every field update
- Cache When Possible: Don't regenerate if inputs haven't changed
- Optimize Prompts: Concise prompts are faster and more cost-effective
- Batch Operations: Use rules for bulk AI generation rather than real-time
Quality Control
- Validation Rules: Ensure AI output meets requirements
- Review Process: Build in human review for critical content
- Fallbacks: Have backup plan if AI generation fails
- Testing: Test with various input scenarios
- Monitoring: Track AI output quality over time
Common Form AI Patterns
Pattern 1: Data Enrichment
User enters minimal data, AI fills in detailed information.
Pattern 2: Content Enhancement
User writes draft, AI improves grammar, style, and completeness.
Pattern 3: Template Filling
AI populates standard template with custom data.
Pattern 4: Multi-Version Generation
AI creates multiple versions for user to choose from.
Pattern 5: Summarization
AI condenses lengthy input into brief summary.
Troubleshooting Form AI
AI Not Generating
- Check field references are correct
- Ensure input fields have data
- Verify AI is enabled on correct plan
- Check for error messages in console
Poor Quality Output
- Make prompt more specific
- Provide more context in prompt
- Add examples of desired output
- Ensure input data is complete
Slow Generation
- Simplify prompt
- Reduce amount of input data
- Use manual vs auto-generation
- Check for unnecessary regeneration
Next Steps
Next: AI in Automation - Using AI in Record Rules, Table Rules, and Action Links
Hands-On Exercise (To Be Added)
Exercise placeholders will include practical activities such as:
- Creating a single-field AI for product descriptions
- Setting up AI prompt buttons for content enhancement
- Building a multi-field AI processor
- Testing different prompt variations and comparing results
Knowledge Check (To Be Added)
Quiz questions will test understanding of:
- Differences between single-field and multi-field AI
- When to use auto-generate vs manual AI
- Best practices for form AI prompts
- Common form AI patterns and use cases
We'd love to hear your feedback.