Alerts
Find out from a dashboard, not a customer.
Set threshold rules on cost, latency, error rate, and eval scores. Foglamp checks them every minute and tells you the moment something drifts.
Error-rate spike3.1%
Daily spend ceilingOK
Thresholds
Rules on the metrics that matter.
Spend over $1k/day. p95 latency past 5 seconds. Error rate above 2%. Groundedness under 0.85. If you can chart it, you can alert on it.
- Cost, latency, error rate, and eval-score rules
- Evaluated every minute on rolling windows
- Scope alerts to an agent, model, or the whole project
Error-rate spike3.1%
Daily spend ceilingOK
Delivery
Routed to where your team already is.
Alerts land in email and Slack the instant they fire, with a direct link to the traces behind the spike, so the on-call path goes straight from notification to root cause.
- Email and Slack delivery
- Deep link from an alert to the offending traces
- Auto-resolves when the metric recovers
Status · last check 28s ago
1
firing
3
healthy
60s
interval
Your agents are running in the fog.
What they cost, when they break, what they say. You can't see any of it. One prompt turns the light on.