Feature
Error Rate Alerts
Get email when failed LLM requests exceed a percentage you set — scoped to the organization, team, user, or key that owns the traffic.
Error-rate threshold and scope before notifications go out.
Signal
Error rate
Threshold
Percent
Scope
Org + user
Threshold
10%
Notify above selected error rate.
Provider
Primary
Route causing failures.
Response
Fallback
Move traffic if needed.
New capabilities
What your team gains with Concentrate
Threshold that means action
Set the error-rate percentage that should trigger an alert, so a normal blip stays quiet and a real failure gets attention.
Scope to the failing surface
Watch error rates by organization, team, user, or key, so the alert points at the specific workload that's degrading.
Straight to the cause
Jump from the alert into request logs to find the model route, provider, or key behind the spike, then reroute or fail over.
Who Concentrate is designed for
See provider failures before every app team does
Error rate alerts watch the share of failed requests for a scope you choose. When failures climb past your threshold, the owner gets notified and can open request logs or shift the route in request routing.
Platform and SRE
Catch a degrading provider or route before it becomes a company-wide incident.
Team owners
Scope alerts to the key or team behind a production workload.
Reliability planning
Pair alerts with fallbacks so notifications lead to a configured backup, not improvisation.
Tunable thresholds
Set the failure percentage that should page someone versus normal retry noise.
Feature basics