01
Pre-incident drift detection
Latency slope, error spread, queue depth, saturation, and deploy recency are scored together. You get paged when the curve starts turning, not when the graph is already on fire.
Wake up to your incidents.
Incident lead time: 08m 42s
Pulse watches deploys, latency, saturation, and error drift together, then pages the right room before the customer-visible blast radius forms.
Signals that refuse to blink.
Pulse treats every service like a pressure vessel. It reads the shape of change, compares it to deploys and dependency behavior, then writes the alert with the culprit already attached.
02
One alert with the evidence
Each alert includes service, suspected deploy hash, related dependencies, current threshold, previous baseline, and suggested owner. No screenshot archaeology at 03:00.
03
Suppression without silence
Pulse dampens noisy repeats while keeping a live paper trail. The room stays quiet until the situation changes or the blast radius expands.
Connect services without ceremony.
Ship metrics, traces, logs, and deploy events through the same collector path. Pulse maps service ownership and dependency direction from the traffic it sees.
Baseline the boring minutes.
The bunker learns normal p99 bands, error scatter, retry storms, and queue behavior per service. The useful alert is the one that knows what yesterday looked like.
Send the room a case file.
When the signal holds, Pulse opens the incident room with cause hints, rollback candidates, owner route, and the exact graph slice that moved first.
Pay for signal, not panic.
Three contracts. Same hard-edged alert engine. Choose by service count, retention window, and the number of rooms you need armed before sunrise.
Hobby
solo$19 / mo
5 watched services
7 day hot telemetry
Email and webhook alerts
Deploy hash correlation
For side projects that still deserve to wake you before users do.
Most popular
Team
on-call$149 / mo
50 watched services
30 day hot telemetry
Slack, PagerDuty, webhook
Owner routing and suppression
For teams shipping often enough that every deploy needs a witness.
Enterprise
bunkerCustom
Unlimited watched services
13 month archive
SAML, audit logs, VPC routes
Incident drills and support desk
For regulated systems where the alert history is part of the evidence chain.
Stay awake.
Wire Pulse into your deploy stream, give it the service map, and let the bunker watch the pressure while your team sleeps.
Install window
First signal in under one hour for a typical HTTP service fleet.
First signal in under one hour for a typical HTTP service fleet.
Routing contract
Alert lands with owner, deploy hash, dependency path, and escalation timer.
Alert lands with owner, deploy hash, dependency path, and escalation timer.