Skip to main content
Cost alerts help you stay informed about your AI spending and catch potential issues before they become problems. Configure alerts to notify you when spending exceeds thresholds, enabling proactive cost management and budget control.

​
Overview

Alerts monitor your AI spending in real-time and send notifications when spending patterns exceed your defined thresholds. You can configure alerts for:
  • Cost thresholds - Get notified when spending exceeds a specific amount
  • Time periods - Monitor spending over different time windows (hour, day, week, month)
  • Specific filters - Focus alerts on particular models, providers, environments, or features
  • Team members - Choose who receives alert notifications
Alert Configuration Interface - A screenshot showing the alert creation form with threshold amount, time period selector, filter options, and recipient selection
Image Placeholder: This should show the alert configuration form with:
  • Alert name field
  • Threshold amount input
  • Time period dropdown (Last hour, 24 hours, 7 days, 30 days)
  • Filter section (models, providers, environments, features)
  • Recipient selection (team members)
  • Alert frequency/throttling settings
  • Enable/disable toggle
  • Save button

​
Alert Types

​
Cost Threshold Alerts

The most common alert type, cost threshold alerts notify you when spending exceeds a specific amount within a defined time period.

Hourly Alerts

Monitor spending over the last hour. Useful for catching sudden spikes in usage.

Daily Alerts

Track spending over the last 24 hours. Perfect for daily budget monitoring.

Weekly Alerts

Monitor spending over the last 7 days. Great for weekly budget reviews.

Monthly Alerts

Track spending over the last 30 days. Essential for monthly budget management.

​
Configuring Alerts

​
Step 1: Set the Threshold

Define the spending amount that triggers the alert. For example:
  • Alert when spending exceeds $1,000 in the last 24 hours
  • Alert when spending exceeds $10,000 in the last 7 days
  • Alert when spending exceeds $50,000 in the last 30 days
Start with conservative thresholds and adjust based on your typical spending patterns. You can always modify alerts after creating them.

​
Step 2: Choose the Time Period

Select the time window for monitoring:
Best for catching immediate spikes in usage. Useful for monitoring production systems in real-time.
Ideal for daily budget monitoring. Helps you stay within daily spending limits.
Perfect for weekly budget reviews. Provides a broader view of spending trends.
Essential for monthly budget management. Helps track longer-term spending patterns.

​
Step 3: Apply Filters (Optional)

Narrow down alerts to specific contexts:
  • By Model - Alert only for specific AI models (e.g., GPT-4, Claude Opus)
  • By Provider - Alert only for specific providers (e.g., OpenAI, Anthropic)
  • By Environment - Alert only for specific environments (e.g., production, staging)
  • By Feature - Alert only for specific product features
  • By Metadata - Alert based on custom metadata values
Filters help you create targeted alerts. For example, you might want a high threshold for production but a lower threshold for development environments.

​
Step 4: Select Recipients

Choose team members who should receive alert notifications:
  • Owners - Organization owners automatically receive all alerts
  • Admins - Admins can be added to specific alerts
  • Members - Team members can be added to alerts they need to monitor
Make sure to add appropriate team members to alerts. Missing critical alerts can lead to unexpected costs.

​
Step 5: Configure Alert Frequency

Control how often alerts are sent to avoid notification fatigue:
  • Immediate - Send alert as soon as threshold is exceeded
  • Throttled - Limit alerts to once per time period (e.g., once per hour, once per day)
  • Minimum Time Between Alerts - Set a minimum time that must pass before the next alert can be sent
Use throttling for high-frequency alerts to avoid overwhelming your team with notifications while still staying informed.

​
Alert History

View a complete history of all alerts that have been triggered: Alert History - A table showing past alerts with timestamp, threshold, actual spending, status, and recipients
Image Placeholder: This should show a table with:
  • Alert name
  • Trigger timestamp
  • Threshold amount
  • Actual spending amount
  • Time period
  • Status (sent, failed, etc.)
  • Recipients
  • Filter details
  • Ability to view alert details
The alert history helps you:
  • Track alert effectiveness - See how often alerts are triggered
  • Identify patterns - Understand when and why spending spikes occur
  • Review alert configuration - Determine if thresholds need adjustment
  • Audit notifications - Verify that alerts are being sent correctly

​
Managing Alerts

​
Enable and Disable

You can enable or disable alerts without deleting them:
  • Disable - Temporarily stop monitoring without losing configuration
  • Enable - Resume monitoring with existing settings
  • Delete - Permanently remove alerts you no longer need

​
Editing Alerts

Update alert configuration at any time:
  • Modify threshold amounts
  • Change time periods
  • Update filters
  • Add or remove recipients
  • Adjust frequency settings
Changes to alert configuration take effect immediately. Historical alert data is preserved.

​
Email Notifications

When an alert is triggered, email notifications are automatically sent to all configured recipients: Alert Email Notification - A screenshot showing an example alert email with spending details, threshold information, and links to view more details
Image Placeholder: This should show an email template with:
  • Alert name
  • Threshold exceeded message
  • Actual spending amount vs. threshold
  • Time period
  • Filter details (if applicable)
  • Link to view details in dashboard
  • Link to view cost breakdown
  • Unsubscribe or manage alerts link
Email notifications include:
  • Alert details - Threshold, actual spending, and time period
  • Context - Filters applied and relevant metadata
  • Quick actions - Links to view details in the dashboard
  • Cost breakdown - Summary of what’s driving the spending

​
Use Cases

Budget Monitoring

Set alerts at 50%, 75%, and 90% of your monthly budget to track spending progress.

Anomaly Detection

Create alerts with low thresholds to catch unexpected spending spikes early.

Environment-Specific

Set different thresholds for production vs. development environments.

Feature Monitoring

Monitor spending for specific high-cost features to ensure they stay within budget.

​
Best Practices

1

Start Conservative

Begin with higher thresholds and gradually lower them as you understand your spending patterns better.
2

Use Multiple Alerts

Set up alerts at different thresholds (e.g., 50%, 75%, 90% of budget) to get early warnings.
3

Filter Appropriately

Use filters to create targeted alerts that focus on what matters most to your team.
4

Review Regularly

Periodically review alert history to ensure thresholds are appropriate and alerts are effective.
5

Throttle High-Frequency Alerts

Use throttling to prevent alert fatigue while still staying informed about important changes.
Pro tip: Combine alerts with Custom Dashboards to create a comprehensive monitoring setup. Use dashboards for regular review and alerts for immediate notifications.
  • Dashboard - Monitor spending trends and patterns
  • Cost Explorer - Investigate spending details when alerts are triggered
  • Reports - Generate detailed reports for post-alert analysis