Relay
Start free
Product How it works Customers Pricing Docs Sign in

Page the right engineer. Let everyone else sleep.

Relay routes production alerts to whoever owns the broken service — by schedule, severity and time of day — and escalates on its own when nobody answers. Built for teams of 3 to 50 engineers.

Start free for 5 engineers Walk through a real incident

No credit card. Imports PagerDuty and Opsgenie schedules in one click.

Used by on-call teams at

Fernhill quillpay TIDEWATER Brightloom kestrel/os Parcelbird
What Relay does

Built for the 3 a.m. version of your team.

Most alerting tools were designed for 400-person ops orgs. Relay is the opposite: three things done carefully, so a small team can carry a pager without burning out.

Routing rules that read like sentences

Write who gets woken up in plain conditions — service, severity, time window, tags from the alert payload. Rules run top to bottom and the first match wins, so you can always answer “why did this page me?”

Routing · production 5 rules · edited 2 days ago by Ana L.
  1. When severity is sev1 → page owning team’s primary by push, SMS and voice; escalate every 5m
  2. When service is marketing-site and time is 22:00–08:00 → hold until 09:00, then notify
  3. When severity is sev2 → page primary by push and SMS; escalate after 15m
  4. When severity is sev3 and time is 08:00–22:00 → post to team’s Slack channel, no page
  5. When severity is sev3 and time is 22:00–08:00 → hold until 09:00 and open a Linear issue

Send a test alert — pick a combination and see which rule catches it.

Service
Severity
Local time
Rule 3 matchedPages Maya Okafor (payments primary) by push and SMS. Escalates to Dev Patel after 15 minutes without an ack.

Schedules that account for people

Weekly rotations, follow-the-sun handoffs across time zones, and overrides that take seconds. Type /relay cover me today 14:00–18:00 in Slack and Relay asks the next person in rotation to accept.

One incident, not forty pages

Relay groups alerts that fire together — same service, same deploy, within a window you choose — and suppresses flapping monitors until they settle. You get woken once, with the full picture.

4,812

alerts sent to Relay by Fernhill in August

311

incidents after grouping and flap suppression

Median engineer at Fernhill: 2 night pages a month, down from 11.

Takes alerts from DatadogGrafanaPrometheus AlertmanagerSentryAWS CloudWatchHoneycombany webhook and talks to SlackLinearGitHubStatuspage
Inside an incident

One screen from first page to postmortem.

This is INC-2041 as Dev saw it at 02:20, thirty seconds after acknowledging. The numbered markers explain what Relay assembled before he even opened his laptop.

relay / payments / incidents / INC-2041
Acknowledged 02:19:41
Sev1INC-2041Acknowledged

checkout-api · p99 latency above 1.2s

Opened 02:14:07 · payments · us-east-1

p99 latency · /v2/charge · last 30 minnow 1,610 ms
threshold 1.2s deploy 4f2c1e
Grouped: p99 latency > 1.2s 5xx rate 4.1% pod restarts ×6
  1. 02:14:073 alerts from Datadog grouped into INC-2041
  2. 02:14:09Paged Maya Okafor push 02:14:09 · SMS 02:14:10 · call 02:15:09, no answer
  3. 02:19:09Escalated to Dev Patel (secondary) call answered 02:19:38
  4. 02:19:41Acknowledged by Dev · Maya’s pages cancelled

Responders

DP
Dev Patelsecondary · ack’d
Lead
MO
Maya Okaforprimary · pages stopped
AL
Ana Lindqvistdeploy author · invited

Likely cause

api@4f2c1e deployed by Ana L. at 02:09, five minutes before the spike.
“Switch charge retries to exponential backoff”
Landed 5m before first alert

War room

#inc-2041-checkout-latency
Created in Slack · 3 members · timeline mirrored

  1. Related alerts, folded together

    Three Datadog monitors fired within 40 seconds. Relay grouped them by service, so Maya got one page instead of three.

  2. Every page and handoff, timestamped

    Push, SMS, then a voice call. When five minutes passed without an ack, Relay called the secondary — and stopped paging Maya the moment Dev answered.

  3. The deploy that probably did it

    Relay reads recent deploys from GitHub and flags the one that landed just before the spike, then invites its author.

  4. A Slack channel, already set up

    Responders are invited and the timeline is mirrored as it happens. When you resolve, the channel becomes the first draft of the postmortem.

Customers

Teams who got their nights back.

“We used to page all six of us for every disk alert. Now the one person who owns that box gets it, and the rest of us hear about it at standup. Nobody has quit over on-call since we switched — that used to be a real conversation.”
11 → 2night pages per engineer, monthly
4m 10smedian time to acknowledge
Priya RamanStaff SRE, Fernhill
“The Slack override is the feature I didn’t know I needed. Swapping a shift for a dentist appointment takes ten seconds and nobody has to edit a calendar.”
Tomás IbarraBackend Lead, Quillpay
“We moved off PagerDuty in an afternoon. Schedules imported, our escalation policies became Relay rules, and the bill for 14 engineers dropped by about two-thirds.”
Hannah LeclercVP Engineering, Tidewater Health
Pricing

Priced per engineer on the rotation.

Only people who can be paged count as seats. Managers, stakeholders and Slack observers are free.

Starter

For a side project or a team’s first rotation.

$0up to 5
engineers

Free forever

Start free
  • 1 schedule, 3 routing rules
  • Push and email notifications
  • Slack app and overrides
  • 7-day incident history
Team Most teams pick this

Everything you need to run on-call properly.

$9per engineer
per month

Billed annually · $11 month-to-month

Start 14-day trial
  • Unlimited schedules and rules
  • SMS and voice calls, 300/month included
  • Escalation chains and follow-the-sun
  • Alert grouping and flap suppression
  • Deploy correlation via GitHub
  • 1-year history and postmortem drafts
Business

For teams with security reviews and an SLA to keep.

$19per engineer
per month

Billed annually · invoicing available

Talk to us
  • Everything in Team
  • SAML SSO and SCIM provisioning
  • Audit log with 3-year retention
  • 1,000 SMS/voice per month
  • 99.99% uptime SLA, 1-hour support response

Compare plans

FeatureStarterTeamBusiness
Engineers on rotationUp to 5UnlimitedUnlimited
Schedules / routing rules1 / 3UnlimitedUnlimited
SMS and voice per month—3001,000
Alert grouping—IncludedIncluded
Deploy correlation—GitHubGitHub, GitLab, Argo CD
Incident history7 days1 year3 years
SSO / SCIM——SAML + SCIM
SupportCommunityEmail, next business day1-hour, 24/7

Your next page should go to one person.

Connect Datadog or Grafana, import your rotation, and send a test alert. Most teams are routing real pages within 20 minutes.