Skip to main content

​
Analytics Overview

Analytics provides comprehensive monitoring and analysis capabilities for your AI agents in a unified dashboard. Track conversations and executions, analyze agent performance with AI-generated insights, and monitor activity in real-time with advanced filtering—all from a single interface.

​
What is Analytics?

Analytics is an integrated dashboard that combines real-time monitoring with AI-powered conversation analysis:

Unified Dashboard

Single interface with conversations and executions in one view

Live Mode

Toggle real-time monitoring to watch activity as it happens

AI-Powered Analysis

Generate conversation insights with AI (up to 250 conversations)

Execution Metrics

Track total executions, success rate, and active runs

​
The Analytics Interface

Analytics provides a unified dashboard with integrated filtering and visualization:
Switch between historical and real-time monitoring
  • OFF: View historical data with date range filters
  • ON: Stream live conversations and executions as they happen
Enable Live Mode when monitoring production, testing new features, or during launch events.
Filter by time periodSelect any date range to analyze:
  • Custom date ranges
  • View historical patterns
  • Compare different periods
  • Focus on specific timeframes
Date range filtering is disabled when Live Mode is active, as live data shows current activity only.
Filter by specific users
  • All Users: View all activity (default)
  • Specific User: Track individual user interactions
  • User Groups: Analyze cohort behavior
Use cases:
  • VIP customer tracking
  • User journey analysis
  • Issue investigation
  • Support requests

​
Activity Timeline

Visual chart showing conversations and executions over time:

Conversations

Black bars showing conversation volume per day

Executions

Teal bars showing execution count per day
Key insights:
  • Identify peak activity periods
  • Spot usage patterns
  • Compare conversations vs executions
  • Detect anomalies or trends

​
Key Capabilities

​
1. Conversations Tab

View and analyze conversation data:
Browse conversationsLeft sidebar shows:
  • Conversation titles (AI-generated)
  • User identification
  • Timestamps
  • Click to view messages
Features:
  • Scroll through filtered conversations
  • Quick conversation access
  • Contextual previews

​
2. Executions Tab

Monitor agent execution performance:

Total Executions

Overall execution countShows:
  • Total executions
  • Completed count
  • Failed count

Success Rate

Execution success percentageTracks:
  • Success percentage
  • Performance trends
  • Quality metrics

Active Executions

Current running executionsDisplays:
  • Currently running
  • Real-time status
  • Active count
Execution List:
  • Each execution shows:
    • Status badge (COMPLETED, FAILED, RUNNING)
    • Execution type (AGENT_RUN)
    • Timestamp
    • “Logs” button for detailed inspection

​
3. Execution Logs

Click any execution’s “Logs” button to view:

Execution Details

  • Step-by-step execution trace
  • Processing timeline
  • Status transitions
  • Error messages (if any)

Tool Interactions

  • Tools called during execution
  • Tool parameters
  • Tool results
  • Execution timing

Performance Data

  • Processing time
  • Token usage
  • API calls
  • Resource consumption

Debug Information

  • Error details
  • Stack traces
  • Context information
  • Debugging aids

​
4. Secure Report Storage

Conversation analyses are automatically stored securely:
1

Run Analysis

Generate AI-powered insights from your filtered conversations
2

Automatic Storage

Analysis results are saved to secure file storage
3

Retrieve Anytime

Access previous analyses from your conversation list
4

Long-Term Access

Reports remain available for historical reference
Security & Privacy: All conversation data and analyses are stored with enterprise-grade security, ensuring data protection and regulatory compliance.

​
Use Cases

​
1. Quality Assurance

Response Quality

  • Review agent responses
  • Identify improvement areas
  • Track quality metrics
  • Ensure consistency

Compliance Monitoring

  • Audit conversations
  • Verify guardrail effectiveness
  • Check policy adherence
  • Generate compliance reports

​
2. Performance Optimization

Identify and resolve performance issues:
  • Analyze slow conversations
  • Find bottlenecks in execution
  • Optimize tool usage
  • Reduce response times
  • Monitor resource consumption

​
3. User Behavior Analysis

Understand how users interact with your agent:
  • Common question patterns
  • Conversation flow analysis
  • User satisfaction indicators
  • Feature usage tracking
  • Drop-off point identification

​
4. Debugging & Troubleshooting

Quickly diagnose and fix issues:
  • Reproduce error scenarios
  • Analyze failed executions
  • Track error patterns
  • Identify edge cases
  • Test fixes with historical data

​
How to Use Analytics

1

Access Analytics Dashboard

Navigate to your agent’s Analytics section from the dashboard
2

Configure Filters

Use the sidebar to set:
  • Live Mode: Toggle ON for real-time, OFF for historical
  • Date Range: Select period to analyze (disabled in Live Mode)
  • User Filter: Choose “All Users” or specific user
Click Filter to apply, Clear to reset
3

View Activity Timeline

Review the chart showing conversations (black) and executions (teal) over time
4

Switch Between Tabs

  • Conversations: View conversations and run AI analysis
  • Executions: Monitor execution metrics and logs
5

Generate AI Insights (Conversations Tab)

Click “Run New Analysis” to generate AI-powered conversation insights
Analysis processes up to 250 conversations from your filtered dataset.
6

Inspect Execution Logs (Executions Tab)

Click “Logs” on any execution to view detailed execution information

​
Important Limitations

Conversation Analysis Limit: 250 ConversationsAI-powered conversation analysis can process up to 250 conversations maximum per analysis run. This ensures:
  • Fast AI processing
  • Optimal performance
  • Reliable insights
  • Quick results
Important: Analysis results are AI-generated estimates and may vary between different analysis runs.
For datasets exceeding 250 conversations, narrow your date range or use user filters to focus the analysis on the most relevant data.

​
Benefits

Visibility

Complete transparency
  • Full conversation history
  • Execution details
  • Real-time updates
  • Historical data access

Insights

Actionable intelligence
  • Usage patterns
  • Performance metrics
  • User behavior
  • Improvement opportunities

Optimization

Continuous improvement
  • Identify bottlenecks
  • Improve responses
  • Optimize costs
  • Enhance user experience

Compliance

Regulatory adherence
  • Audit trails
  • Data retention
  • Access logs
  • Compliance reports

Debugging

Faster problem resolution
  • Error tracking
  • Root cause analysis
  • Reproduction scenarios
  • Fix validation

Security

Protected data
  • Secure storage
  • Encrypted reports
  • Access control
  • Privacy compliance

​
Live Mode vs. Historical Mode

Toggle between two modes using the Live Mode switch:
Real-time activity monitoring✅ Best for:
  • Production monitoring
  • Launch events
  • Testing new features
  • Immediate issue detection
  • Current system health
⚡ How it works:
  • Streams live updates automatically
  • Shows current conversations and executions
  • Date range filter is disabled
  • Updates appear instantly
  • No refresh needed

​
Next Steps

Conversation Analysis

Learn how to generate AI-powered conversation insights

Live Mode

Discover real-time monitoring with Live Mode

Agent Settings

Configure your agent for optimal analytics

API Reference

Explore the Analytics API

​
Frequently Asked Questions

The 250 conversation limit ensures fast AI processing and optimal performance. For larger datasets, narrow your date range or use user filters to focus on the most relevant conversations.
Live Mode ON: Streams real-time activity as it happens, perfect for monitoring production. Live Mode OFF: Enables historical analysis with date range filtering and AI-powered conversation insights.
Yes, you can combine date range AND user filters to create highly targeted analysis. For example, track a specific user’s activity during a particular time period.
No, the conversation analysis note states: “Analysis results are estimates generated by AI and may vary between different analysis runs.” This is because AI models can produce different insights on each run.
Switch to the Executions tab and click the “Logs” button next to any execution to view detailed information including execution traces, tool calls, and performance metrics.
Yes, in the Conversations tab, click on any conversation in the left sidebar to view the complete message history with all user inputs and agent responses.
The Executions tab displays three key metrics: Total Executions (completed and failed count), Success Rate (percentage), and Active Executions (currently running).
Yes, all conversation data and AI-generated analyses are stored with enterprise-grade security and encryption to ensure data protection and regulatory compliance.