Monitoring Agents

Once your agents are running, you'll want to keep an eye on them. This guide covers how to monitor performance, view logs, and troubleshoot problems.

Agent Dashboard

Access your agent overview:

1. Click "Agents" in the sidebar
2. See all agents with status indicators:
   • Active (green) - Running normally
   • Paused (yellow) - Temporarily disabled
   • Error (red) - Last run failed
   • Draft (gray) - Not yet activated

Viewing Run History

Understanding Run Status

Each run has a status:

Execution Logs

Every run produces detailed logs showing:

Step-by-Step Execution

[2024-01-15 09:00:01] Agent started
[2024-01-15 09:00:02] Reading from Gmail...
[2024-01-15 09:00:05] Found 12 unread emails
[2024-01-15 09:00:06] Processing emails...
[2024-01-15 09:00:15] Generated summary
[2024-01-15 09:00:16] Sending to Slack #daily-digest
[2024-01-15 09:00:17] Message sent successfully
[2024-01-15 09:00:17] Agent completed

Tool Calls

See every action the agent took:

• Which integrations were accessed
• What data was read/written
• API responses received

AI Reasoning

View the model's thinking:

• How it interpreted instructions
• Decisions it made
• Why it took certain actions

Notifications

Email Notifications

Receive emails when:

• Agent completes (if configured)
• Agent fails
• Agent requires attention

Configure in agent settings → Notifications.

In-App Notifications

See alerts in Billix:

• Bell icon shows recent notifications
• Click to see details
• Mark as read or take action

Setting Up Alerts

Failure Alerts

Get notified immediately when an agent fails:

1. Open agent settings
2. Go to Notifications
3. Enable "Alert on failure"
4. Choose notification method (email, Slack, etc.)

Performance Alerts

Monitor for slow runs:

1. Open agent settings
2. Set expected completion time
3. Alert if run exceeds threshold

Debugging Failed Runs

When an agent fails, follow these steps:

Common Issues and Solutions

Authentication Errors

Symptom: "Failed to authenticate with Gmail"

Solutions:

• Go to Settings → Integrations
• Disconnect and reconnect the service
• Ensure you granted all required permissions
• Check if your password changed

Timeout Errors

Symptom: "Agent exceeded maximum execution time"

Solutions:

• Simplify instructions
• Process fewer items per run
• Break into multiple agents
• Use a faster AI model

Rate Limit Errors

Symptom: "Rate limit exceeded for Gmail API"

Solutions:

• Reduce run frequency
• Process fewer items
• Add delays between API calls
• Contact support for limit increase

Not Found Errors

Symptom: "Channel #daily-updates not found"

Solutions:

• Verify the resource exists
• Check for typos in names/IDs
• Ensure agent has access permissions
• Update configuration if renamed

Performance Metrics

Track agent performance over time:

Execution Time

• Average run duration
• Trend over last 30 days
• Comparison to baseline

Success Rate

• Percentage of successful runs
• Failure patterns
• Common error types

Resource Usage

• Credits consumed per run
• Total credits this month
• Model usage breakdown

Best Practices for Monitoring

Regular Check-ins

• Review agent dashboard weekly
• Check for new failures
• Monitor success rates

Set Up Alerts

• Enable failure notifications
• Consider success notifications for critical agents
• Set up Slack alerts for team visibility

Log Retention

Export important logs before they expire.

Document Issues

Keep notes on:

• Recurring problems
• Solutions that worked
• Configuration changes

Common Questions

Next Steps

• Review best practices for reliable agents
• Learn about triggers for scheduling
• Explore templates for proven patterns