Checked from 30 cities. Every 30 seconds.
API errors in Singapore and Hong Kong
41 alerts grouped · Mira paged
Watch every endpoint from where users are.
A check fails only when two regions agree. Then a third region retries before anyone is paged.
- Check types
- HTTP and API assertions
- Multi-step browser flows
- SSL and domain expiry
- Cron heartbeats
| Monitor | Uptime |
|---|---|
| API healthapi.halden.dev/v1/health | 99.98% |
| Checkout flowhalden.dev/checkout | 99.95% |
| Primary databasedb-1.halden.internal:5432 | 100% |
| Nightly exportcron · every day 03:00 | 99.2% |
| TLS certificatehalden.dev · 41 days left | 100% |
Page one person, not the whole team.
The assistant reads every alert, matches region, endpoint and the last deploy, and opens a single incident.
- Included
- Rotations and overrides
- Escalation policies
- Phone, SMS, Slack and push
- Quiet hours per person
Rotations
Weekly or follow-the-sun. Swap a shift from your phone.
Escalation
If nobody acknowledges, the next step runs on its own.
- 0 minPush and Slack to on-call
- 3 minPhone call to on-call
- 8 minPhone call to backup
- 15 minWhole platform team
Alert grouping
Forty alerts about one cause become one incident.
APAC API errors after deploy 4821
Every incident writes its own timeline.
Checks, pages, deploys and status posts land in one log. Export it as the first draft of your postmortem.
- Checks fail in Singapore and Hong Kong503 from 2 of 30 regions, retried from Tokyo.
- 41 alerts grouped into one incidentMatched on region, endpoint and deploy 4821 at 02:11.
- Mira Okafor pagedPhone and Slack.
Acknowledged at 02:16 - Status page updatedDrafted by the assistant, edited by Mira.
- Deploy 4821 rolled backFrom GitHub, marked on the latency chart.
- ResolvedAll 30 regions green for 5 minutes.
A public page that keeps up with you.
Components turn amber the moment a check fails. You decide what customers read.
Monitors in a file. Reviewed like code.
Keep checks next to the service they watch. The CLI shows a diff before anything changes.
import { monitor, regions } from 'pingvale' export default monitor({ name: 'API health', url: 'https://api.halden.dev/v1/health', every: '30s', regions: regions.all(), failWhen: { regionsAgree: 2 }, expect: { status: 200, p95: '<400ms' }, page: 'platform-oncall', statusPage: 'API', })
# pingvale.yaml monitors: - name: API health url: https://api.halden.dev/v1/health every: 30s regions: all fail_when: regions_agree: 2 expect: status: 200 p95: "<400ms" page: platform-oncall
resource "pingvale_monitor" "api" { name = "API health" url = "https://api.halden.dev/v1/health" every = "30s" regions = ["all"] fail_when { regions_agree = 2 } page = pingvale_schedule.platform.id }
$ pingvale deployReading pingvale.ts … 64 monitors + checkout-flow browser 1m ~ api-health every 60s → 30s + db-primary tcp 30s Apply 3 changes? yes✓ Deployed in 2.1 s✓ First checks from 30 regions$ ▍
Works with the tools you page with.
Deploy markers from your CI, alerts where your team already talks.
Fewer pages. Better ones.
“Last quarter we had 1,200 alerts. Pingvale turned them into 38 incidents, and nobody on the team muted their phone.”

“The status update is drafted before I’ve opened my laptop. I change one line and post it.”

“We moved 140 monitors into one file in an afternoon. Changes go through pull requests now.”

Start free. Pay when you add the team.
Flat price per workspace. Phone calls and SMS included up to fair use.
Team
- 100 monitors, every 30 s
- All 30 regions
- On-call for 10 people
- Assistant: grouping and drafts
Scale
- 500 monitors, every 10 s
- Private status pages, SSO
- 13 months of history
- 99.99% uptime SLA
Questions we hear on calls.
How do you avoid false alarms?
A monitor fails only when at least two regions agree, and you can raise that number. Every failure is retried from a third region before anyone is paged.
What does the assistant do, exactly?
It groups alerts that share a region, an endpoint or a recent deploy into one incident, keeps the timeline summary current and drafts status updates. It never posts an update or closes an incident on its own.
Can I keep my current on-call tool?
Yes. Send incidents to it with a webhook and use Pingvale only for checks and status pages.
Where do the checks run?
30 regions on six continents. Where we can, each region runs on two cloud providers, so one provider’s bad day does not look like yours.
How do I move my monitors over?
Import a CSV or an OpenAPI file, or describe monitors in pingvale.ts and run pingvale deploy.
Is there a free plan?
Yes: 10 monitors with 3-minute checks from 5 regions and one status page. No card needed.
Sleep through the nights nothing breaks.
We’re opening Pingvale to teams in batches. Leave your work email and we’ll send an invite.