Skip to main content

Conversation History

View, filter, and analyze all historical conversations with your Agentforce agents.

Conversation History

The Problem​

Understanding how your agents are performing requires visibility into past conversations, including successful interactions and errors.

Teams need to:

  • 📊 Monitor Quality: Review conversations to ensure agents are responding appropriately
  • 🐛 Debug Issues: Investigate specific conversations where users reported problems
  • 📈 Analyze Patterns: Identify common queries, popular topics, and frequent errors
  • 🔍 Filter Effectively: Find specific conversations by date, errors, or content
  • ✅ Audit Behavior: Review agent decisions for compliance and governance
  • 📤 Share Insights: Export conversations for team review or stakeholder updates

In short: You need a centralized view of all conversations to monitor, debug, and improve your agent implementations.

How GenAI Explorer Solves This​

GenAI Explorer provides comprehensive conversation analytics with:

✅ Complete Conversation View: See all sessions, messages, topics, and actions in one place

✅ Advanced Filtering: Find exactly what you need

  • Text search across all content
  • Date range filters (From/To)
  • Error-only conversations
  • By topic or agent
  • Clear all filters with one click

✅ Visual Filter Badges: Active filters shown as colored badges for clarity

✅ Expanded Details: Click any conversation to see

  • Full message timeline
  • Topics invoked
  • Actions executed (success/failure)
  • Step-by-step processing
  • Error messages with context

✅ Analytics & Insights: Built-in metrics

  • Total conversations, interactions, messages
  • Average turns per session
  • Failed action rates
  • Performance trends

✅ Export Capabilities: Export to JSON, CSV, or Text for reporting

Impact: Review quality in minutes, debug issues 5x faster, and gain actionable insights into agent performance and user patterns.

Overview​

The Conversation History feature provides a comprehensive view of all past interactions with your Agentforce agents. It's an essential tool for monitoring agent performance, debugging issues, analyzing user patterns, and ensuring quality customer service.

Key Features​

📊 Complete Conversation View​

  • All Sessions: See every conversation across all agents
  • Message Details: Full transcript with timestamps
  • Topic Information: Which topics were invoked
  • Action Tracking: What actions were executed
  • Error Visibility: Failed actions and error messages

🔍 Advanced Filtering​

Conversation History Filters

  • Text Search: Search across all conversation content
  • Date Range: Filter by time period (From/To dates)
  • Error Filter: Show only conversations with errors
  • Topic Filter: Filter by specific topics
  • Agent Filter: Filter by specific agents/planners

📈 Analytics & Insights​

  • Total Statistics: Messages, interactions, and sessions count
  • Performance Metrics: Response times and success rates
  • Usage Patterns: Most active topics and actions
  • Error Analysis: Failure rates and common issues

Accessing Conversation History​

From the Home Page​

  1. Open GenAI Explorer
  2. Navigate to Home or Agentforce
  3. Look for the "Conversation History" section
  4. Browse the list of conversations

Main View:

  • Home → Conversation History - All conversations across all agents
  • Agent Details → Conversations - Conversations for specific agent
  • Atlas → Query Lab → Recent Requests - Technical request view

Using Filters​

Text Search​

Search across:

  • User messages
  • Agent responses
  • Topic names
  • Action names
  • Error messages

How to use:

  1. Type your search term in the search box
  2. Results filter automatically as you type
  3. Search is case-insensitive
  4. Supports partial matches

Example Searches:

  • "order" - Find all conversations about orders
  • "error" - Find conversations with errors
  • "refund" - Find refund-related conversations

Date Range Filter​

Filter by conversation date:

┌─────────────────────────────────────────┐
│ From: [2024-01-01]  To: [2024-01-31]   │
└─────────────────────────────────────────┘

Options:

  • From Date Only: Show conversations from this date forward
  • To Date Only: Show conversations up to this date
  • Both Dates: Show conversations in the date range

Date Handling:

  • From Date: Includes entire day (00:00:00 start)
  • To Date: Includes entire day (23:59:59 end)
  • Dates are inclusive

Use Cases:

  • Monthly reviews: Set first and last day of month
  • Incident investigation: Set date range around incident
  • Performance analysis: Compare different time periods

Error Filter​

Show only conversations with errors:

☑ī¸ Show only conversations with errors

What counts as an error:

  • Failed action executions
  • API errors
  • Timeout errors
  • Data validation errors
  • Permission errors

When to use:

  1. Troubleshooting: Find problematic conversations quickly
  2. Quality Assurance: Review all failures
  3. Pattern Detection: Identify recurring issues
  4. Root Cause Analysis: Investigate error causes

Topic Filter​

Filter by specific topics:

Topic: [Select Topic â–ŧ]
       - Billing Inquiries
       - Technical Support
       - Account Management
       - Order Status

Use Cases:

  • Analyze specific topic performance
  • Review topic-specific conversations
  • Compare topics against each other
  • Identify topic-specific issues

Agent/Planner Filter​

Filter by agent:

Agent: [Select Agent â–ŧ]
       - Service Agent v2
       - Sales Agent v1
       - Support Agent v3

Use Cases:

  • Compare agent versions
  • Test new agent configurations
  • Isolate agent-specific issues
  • Agent performance analysis

Clear All Filters​

Reset all filters at once:

[🗑ī¸ Clear All Filters]

Clears:

  • Text search
  • Date range (From/To)
  • Error filter checkbox
  • Topic selection
  • Agent/Planner selection

Filter Badges​

Active filters display as badges in the header:

┌─────────────────────────────────────────────────────┐
│ Conversation History                                 │
│                                                      │
│ 🔴 With Errors  đŸ”ĩ Jan 1 - Jan 31  đŸŸĸ Billing      │
└─────────────────────────────────────────────────────┘

Badge Colors:

  • 🔴 Red: Error filter active
  • đŸ”ĩ Blue: Date range active
  • đŸŸĸ Green: Topic/Agent filter active

Click any badge to remove that filter.

Conversation Details​

Expanded View​

Click on any conversation to expand and see:

Conversation Expanded

1. Header Information:

  • Conversation ID
  • Start and end timestamps
  • Duration
  • Agent name and version

2. Statistics:

  • Total messages exchanged
  • Number of topics invoked
  • Actions executed
  • Steps taken (Action/Topic steps)

3. Topics Used:

Topics:
â€ĸ Billing Inquiries
â€ĸ Account Management

4. Actions Executed:

Actions:
â€ĸ GetOrderStatus ✓
â€ĸ ProcessRefund ❌ (Failed)
â€ĸ SendConfirmationEmail ✓

5. Message Timeline:

┌─────────────────────────────────────────┐
│ 👤 USER (2:30 PM)                       │
│ I need help with my order               │
└─────────────────────────────────────────┘

┌─────────────────────────────────────────┐
│ 🤖 AGENT (2:30 PM)                      │
│ I'll help you with your order. Let me   │
│ check the status for you.               │
└─────────────────────────────────────────┘

┌─────────────────────────────────────────┐
│ 🤖 AGENT (2:30 PM)                      │
│ Your order #12345 is in transit...      │
└─────────────────────────────────────────┘

6. Step Details:

Step 1: TOPIC_MATCH → Billing Inquiries (1.2s)
Step 2: ACTION → GetOrderStatus (0.8s)
Step 3: TOPIC_MATCH → Order Status (0.5s)
Step 4: GENERATION → Response Generated (2.1s)

Viewing Options​

Expand/Collapse Controls​

Expand All Button:

[âŦ‡ī¸ Expand All]

Opens all conversations at once to see full details.

Individual Expansion: Click on any conversation header to toggle its expansion.

Display Options​

Markdown Rendering:

[📝 Markdown] Toggle

Enable/disable markdown formatting in agent responses.

Prompt Info:

[ℹī¸ Prompt Info] Toggle

Show/hide system prompts and instructions used.

Refresh:

[🔄 Refresh All]

Reload conversations and get latest data.

Analytics & Insights​

Summary Statistics​

At the top of the page:

┌─────────────────────────────────────────────────────┐
│ 200 total messages â€ĸ 103 interactions â€ĸ 66 sessions │
└─────────────────────────────────────────────────────┘

After Filtering:

┌─────────────────────────────────────────────────────┐
│ Showing 15 conversations matching your filters       │
│ 45 messages â€ĸ 23 interactions â€ĸ 15 sessions         │
└─────────────────────────────────────────────────────┘

Performance Metrics​

Available at bottom of page:

┌──────────────────────────────────────┐
│ 📊 Analytics                         │
│                                      │
│ Total Sessions:        66            │
│ Total Interactions:    103           │
│ Avg Turns/Session:     1.6           │
│ Failed Actions:        19 (18%)      │
└──────────────────────────────────────┘

Use Cases​

1. Quality Assurance Review​

Goal: Review recent agent conversations for quality

Steps:

  1. Set date range to last 7 days
  2. Browse conversations
  3. Check agent response quality
  4. Look for incorrect information
  5. Identify training needs

2. Error Investigation​

Goal: Debug why certain conversations fail

Steps:

  1. Enable "With Errors" filter
  2. Review failed conversations
  3. Check error messages in steps
  4. Identify patterns (same action failing?)
  5. Fix configuration or data issues

3. Topic Performance Analysis​

Goal: Understand how well a topic is performing

Steps:

  1. Filter by specific topic
  2. Review all conversations for that topic
  3. Check success rate
  4. Analyze response times
  5. Identify improvement opportunities

4. Agent Comparison​

Goal: Compare two agent versions

Steps:

  1. Filter by Agent v1, note metrics
  2. Filter by Agent v2, note metrics
  3. Compare:
    • Success rates
    • Response times
    • Error rates
    • User satisfaction patterns

5. Usage Pattern Analysis​

Goal: Understand when and how agents are used

Steps:

  1. Review conversations across date ranges
  2. Identify peak usage times
  3. Common topics and questions
  4. Seasonal patterns
  5. Capacity planning insights

Advanced Filtering Combinations​

Example: Recent Billing Errors​

Filters:
✅ Date Range: Last 7 days
✅ Topic: Billing Inquiries
✅ With Errors: Yes

Result: All failed billing conversations in the past week

Example: Successful Sales Conversations This Month​

Filters:
✅ Date Range: This month
✅ Agent: Sales Agent v2
❌ With Errors: No (unchecked)
✅ Search: "purchase" OR "buy"

Result: All successful sales-related conversations

Example: High-Value Customer Interactions​

Filters:
✅ Search: "premium" OR "vip"
✅ Date Range: Last 30 days
❌ With Errors: No

Result: All premium customer conversations without issues

Exporting Conversations​

Export Options​

1. Individual Conversation:

  • Click conversation
  • Click export icon
  • Choose format: JSON, CSV, or Text

2. Filtered Results:

  • Apply your filters
  • Click "Export Filtered"
  • All matching conversations export

3. Full History:

  • No filters applied
  • Click "Export All"
  • Complete conversation history

Export Formats​

JSON:

{
  "sessionId": "a4j5g000...",
  "startTime": "2024-01-15T14:30:00Z",
  "messages": [...],
  "interactions": [...],
  "steps": [...]
}

CSV:

SessionID,Timestamp,Role,Message,Topic,Action,Error
a4j5g...,2024-01-15 14:30,USER,"Help with order",,
a4j5g...,2024-01-15 14:30,AGENT,"Let me check",Billing,GetOrder,

Text:

Conversation: a4j5g000...
Date: January 15, 2024 2:30 PM

USER: Help with my order
AGENT: Let me check your order status...

Best Practices​

1. Regular Review​

✅ DO:

  • Review conversations weekly
  • Check error rates regularly
  • Monitor topic performance
  • Track improvement trends

❌ DON'T:

  • Wait for issues to escalate
  • Ignore error patterns
  • Skip regular audits

2. Effective Filtering​

✅ DO:

  • Start broad, then narrow
  • Use date ranges for context
  • Combine multiple filters
  • Save common filter combinations

❌ DON'T:

  • Apply too many filters at once
  • Filter so narrowly you miss patterns
  • Forget to clear filters between searches

3. Error Analysis​

✅ DO:

  • Use error filter first
  • Group errors by type
  • Track resolution status
  • Document fixes applied

❌ DON'T:

  • Ignore low-frequency errors
  • Assume one-off errors
  • Skip root cause analysis

4. Performance Monitoring​

✅ DO:

  • Track metrics over time
  • Compare periods
  • Set benchmarks
  • Alert on degradation

❌ DON'T:

  • React to single conversations
  • Ignore seasonal patterns
  • Skip context analysis

Troubleshooting​

Issue: No Conversations Showing​

Possible Causes:

  • Filters too restrictive
  • No data synced yet
  • Date range excludes all data
  • Permission issues

Solutions:

  1. Click "Clear All Filters"
  2. Check date range is reasonable
  3. Wait for Data Cloud sync (5 min)
  4. Verify Data Cloud access

Issue: Missing Message Content​

Possible Causes:

  • PII masking enabled
  • Data not fully synced
  • Permissions limited

Solutions:

  1. Check PII settings
  2. Wait for full sync
  3. Verify field-level security

Issue: Slow Loading​

Possible Causes:

  • Too many conversations
  • No date range filter
  • Large message content

Solutions:

  1. Add date range filter
  2. Limit to last 30 days
  3. Use search to narrow results
  4. Export for offline analysis

Next Steps​