Setting Up AI Agents
Create specialized AI agents that understand your business context and can query your data intelligently.
🤖 What Are AI Agents?
AI Agents are specialized assistants configured to:
- Query specific data sources
- Understand business context
- Follow security rules
- Generate optimized SQL
- Provide domain-specific insights
🎯 Pre-configured Agent Types
Datix xAgents comes with templates for common use cases:
📊 General Data Analyst
Best for ad-hoc analysis and exploration
📈 Sales Performance Analyst
Track sales metrics, revenue, and pipeline
🎯 Marketing Campaign Analyst
Campaign performance and attribution
💰 Financial Data Analyst
Financial reporting and budget analysis
⚙️ Operations Analyst
Process efficiency and metrics
📱 Product Analytics Specialist
User behavior and feature adoption
📝 Creating an Agent
Step 1: Navigate to Agents
- Click "Agents" in the sidebar
- Click "+ New Agent" button
Step 2: Basic Information
Agent Name: Sales Performance Agent
Description: Analyzes sales data, tracks revenue,
and provides insights on customer acquisition
and deal pipeline.
Agent Type: Sales Performance Analyst
Step 3: Select Data Source
Choose which database this agent can query. An agent can only query ONE data source.
Step 4: Configure System Prompt
The system prompt gives the agent business context:
Example for Sales Agent:
You are a sales performance analyst. When analyzing data:
1. Focus on key metrics: revenue, deal size, win rate, pipeline value
2. Always include time comparisons (MoM, YoY)
3. Break down by: region, product, sales rep, customer segment
4. Look for trends and anomalies
5. Suggest actionable insights
Business context:
- Our fiscal year starts in January
- We have 3 product lines: Enterprise, Professional, Starter
- Sales regions: North America, EMEA, APAC
- Average sales cycle: 45 days
Step 5: Table Selection
Choose which tables this agent can access:
- ✅ customers
- ✅ orders
- ✅ order_items
- ✅ products
- ❌ internal_admin (exclude sensitive tables)
Step 6: Column Permissions
Set column-level access control:
- ✅ customer_id
- ✅ customer_name
- ✅ company
- ✅ industry
- ❌ ssn (hidden)
- ❌ credit_card (hidden)
Step 7: Status & Save
- Active: Agent is available for use
- Inactive: Agent exists but can't be used
Click "Create Agent" to save.
👥 Assigning Agents to Users
Admins can control who can use which agents:
Assign to Specific Users
- Go to Users page
- Click on a user
- Click "Manage Agent Assignments"
- Select agents this user can access
- Save
- Control data access by user
- Different agents for different teams
- Maintain security boundaries
⚙️ Agent Configuration Best Practices
✅ Specific System Prompts
Include business context, metrics, and definitions
✅ Limit Table Access
Only grant access to relevant tables
✅ Hide Sensitive Columns
Use column permissions to hide PII and sensitive data
✅ Descriptive Names
Use clear, descriptive agent names
🔄 Managing Agents
Edit Agent
- Go to Agents
- Click on an agent
- Click "Edit"
- Update settings
- Save
Deactivate Agent
Temporarily disable without deleting:
- Edit the agent
- Set Status to Inactive
- Save
Clone Agent
Create a copy for different teams:
- Click agent
- Click "Clone"
- Modify name and settings
- Save
🔐 Security Considerations
Only grant necessary access:
- Minimum required tables
- Minimum required columns
- Role-appropriate permissions