Using AI Assistant for Data Mapping

Overview

When to Use AI Assistant

• You have many tables/collections and want to focus on specific ones
• You're not sure which entities are relevant for your use case
• You want to understand relationships between data
• You need help identifying business entities

Step 1: Access Data Mapping

1. In Agent Builder, after Schema Analysis, you'll reach Data Mapping
2. You'll see the Entity Relationship graph showing all your data
3. Look for the AI Assistant box at the top

Step 2: Use Natural Language Instructions

• "I want to create agents to show all the outlets and sales data only"
• "Map those data for orders only"
• "Show me products and inventory information"
• "Include customer and order data"

Step 3: Review AI Suggestions

1. Filter entities: Show only relevant tables/collections
2. Suggest related entities: Recommend additional entities you might need
3. Explain reasoning: Tell you why it selected certain entities

Step 4: Iterate and Refine

• Add more filters: "Also include employee data"
• Remove entities: "Remove inventory tables"
• Refine focus: "Focus only on sales transactions"

Step 5: Review the Graph

1. The Entity Relationship graph updates automatically
2. You'll see:
3. The graph fits to show only relevant entities

• - Entities: Number of tables/collections shown
• - Relationships: Connections between entities
• - Business Entities: Mapped business concepts

Step 6: Reset if Needed

• Click Reset to Complete View to see all entities again
• This clears all filters and shows the full schema

AI Assistant Features

Smart Entity Recognition

• "sales" → maps to orders, transactions, invoices
• "products" → maps to products, items, inventory
• "customers" → maps to customers, clients, users

Relationship Detection

• Foreign key relationships
• Many-to-many relationships
• Hierarchical structures

Context Awareness

• Your agent type
• Your data source structure
• Common business patterns

Best Practices

1. Be Specific: "Show sales data" is better than "show data"
2. Use Business Terms: Use terms you understand, not technical table names
3. Iterate: Start broad, then narrow down
4. Review Suggestions: The AI's suggestions are helpful but review them
5. Reset When Needed: Don't hesitate to reset and start over

Example Workflow

1. Initial Prompt: "I want to analyze sales data"

• - AI shows: orders, order_items, products, customers

1. Refine: "Focus on orders and products only"

• - AI filters to: orders, order_items, products

1. Add Context: "Also include outlet information"

• - AI adds: outlets table

1. Final Review: Check the graph shows the right entities

Troubleshooting

• Try rephrasing your request
• Be more specific about what you need
• Use business terms instead of technical names

• Wait a moment for processing
• Try clicking Reset and starting over
• Check browser console for errors

• Use more specific instructions
• Add or remove filters iteratively
• Review AI suggestions carefully

Advanced Tips

• Combine multiple filters: "Show sales and inventory, but exclude employee data"
• Use negation: "Everything except financial records"
• Reference by function: "Data needed for order processing"
• Specify relationships: "Include all tables related to customers"