Skip to content

Status pages thatread the metrics,not the operator.

Observer evaluates the same metrics your engineers look at, then publishes the verdict. One backend. One source of truth. A 99.99% contract and a 99.5% contract from the same probe.

Free tier, no card,
first metric public in ~10 minutes

API availabilityAPI latency (p95)API error rate
Receiver latency (p95)Ingest rejections (rate)Agents onlineAgents lagging
Database connectivityDB query latency (p95)Replication lagWAL archive failures (1h)
Queue backlogOldest queued message
Webhook dead letters (1h)Webhook delivery (p95)
Email / notification errors
Status pageDocs siteConsoleMCP availabilityMCP latency (p95)MCP error rate

The same incident, on two status pages.

The checkout API begins returning 5xx at 10:14. One page reads what the operator typed, the other reads what the metric measured.

Traditional, operator-driven
no incident00:00
Observer, metric-driven
watching00:00
All services operational
servicesall within threshold
Checkout API5 metrics
0.08%
Webhooks3 metrics
0.10%
incident logno incidents
No thresholds crossed. 47 metrics polled, all green.
10:1010:1510:2010:2510:3010:3510:4010:4510:5010:5511:00

Four places every status page leaks.

Operator memory, page-shaped services, outside-in probes, one threshold for every customer. Each is the moment your status page stops telling the truth. Here is how each gets fixed.

Memory1 / 4
metric: checkout.p99
type: prometheus
query: |
histogram_quantile(0.99,
sum by (le)(rate(http_req[5m])))
thresholds:
degraded: > 250ms
unhealthy: > 500ms
slo: 99.95%
customers: [mercury, helios]

How it works.

From the metrics you already watch to a published verdict — three stages, one network boundary.

01Sources

Your signals

PrometheusOpenTelemetryCloudWatchPostgres / MySQL / Mongo / RedisLoki / ElasticsearchHTTP / TCP / DNS / TLS / gRPC probes

The metrics your engineers already watch.

read-only
02Boundary

The agent (running locally)

your network
observer-agent

raw metric data stays here

Open-source, outbound-only. The evaluation happens inside your network; only the conclusion leaves it.

{ metric_id, value, status, timestamp }the only payload that crossesNot raw time-series, not query results, not log lines.
03Publish

Verdicts, published

status page

All systems operational

notifications

Slack · Telegram · Discord · Teams · PagerDuty · Splunk On-Call · Grafana IRM · Jira · Webhooks · RSS

The page flips in seconds; recovery flips it back; nobody types anything.

See a real page.

This is not a mockup. status.use.observer runs on Observer, reading Observer's own production metrics right now. When Observer degrades, you see it here first.

status.use.observer

Live production page · auto-published verdicts · zero manual updates

Why teams switch.

Every signal your engineers already watch, a verdict per customer, and a human between detection and publish. The same metric that triggers your on-call also flips your public page.

Engineering signals

Prometheus, OpenTelemetry, CloudWatch, read-only queries on Postgres, MySQL, MongoDB and Redis, log aggregations from Loki and Elasticsearch — plus HTTP, TCP, DNS, TLS, ICMP (needs raw-socket privileges on the agent host), gRPC and WebSocket probes and built-in host metrics. One evaluator behind all of them: the metric on your dashboard is the metric on your status page.

Customer-scoped verdicts

A 99.99% contract and a 99.5% contract from the same probe. The same 247 ms reading is healthy for the customer on 99.5% and unhealthy for the one on 99.99% — each sees their own page, their own SLO, their own incidents, all computed from a single metric.

Human-approved automation

Burn-rate alerts fire when the error budget drains too fast. Incidents auto-draft the moment a threshold is crossed and email your on-call a one-click publish — nothing reaches customers without approval, and drafts nobody acts on expire after 24 hours. Once published, updates fan out to Slack, Teams, PagerDuty, webhooks and RSS.

The agent stays in your network

The Observer agent runs inside your network, outbound-only, and evaluates every metric locally. A verdict — the computed healthy, degraded, or unhealthy state — is all that crosses to the cloud. Queries, credentials, and raw telemetry never leave. Apache-2.0; audit every line.

AI & API native

A public REST API — 27 operations, OpenAPI spec included. Export and apply pages, services, metrics and SLOs as observer/v1 YAML: PR-reviewed, reproducible, reversible, no clickops between you and production. Tell Claude to 'set up Observer for checkout' and the MCP server does it end-to-end — Claude and Codex skills, llms.txt for the docs.

Flat pricing

No per-seat pricing. Add as many observers as you want — the team using the tool most isn't punished for using it. $0 free, $19 starter, $79 pro, $499 enterprise. Customer-scoped pages from $79, not $1,500+ enterprise.

The agent runs inside your network and is outbound-only — raw metric data never crosses into Observer Cloud, only the precomputed verdict and status events do. Read more

Primary in Europe, continuously-replicated disaster-recovery replicas on separate providers — and the agent buffers verdicts locally through any cloud unavailability, so your history has no gap. Read more

The agent is open source under Apache-2.0 — read exactly what it does, audit the network behaviour, and build it yourself. Read more

Pricing

Four tiers. Pick the one whose quotas you'll actually hit. Switch up or down at any time.

Free

Single status page, single agent. Get your first metric to public in 10 minutes.

$0/month

Start free
What's included

1 status pages

1 agents

10 metric definitions

5 email subscribers

7d metric history

Community support

Starter

For a single product launching to its first paying customers. All integrations included.

$19/month

Billed monthly, cancel any time, taxes may apply.

Start Starter
What's included

1 status pages

3 agents

50 metric definitions

3 SLOs

1 custom domains

5,000 email subscribers

30d metric history

Customer-scoped pages

SLA reports per customer

Executive dashboard

Compliance evidence pack

Outbound webhooks

Auto-incident creation

Custom CSS

Standard support

ProMost popular

For SaaS that already has customers. Customer-scoped pages, more SLOs, more domains.

$79/month

Billed monthly, cancel any time, taxes may apply.

Start Pro
What's included

3 status pages

10 agents

500 metric definitions

Unlimited SLOs

5 custom domains

50,000 email subscribers

90d metric history

Customer-scoped pages

SLA reports per customer

Executive dashboard

Compliance evidence pack

Outbound webhooks

Auto-incident creation

Custom CSS

Standard support

Business

For B2B SaaS scaling past 25 customers. 100 customer-scoped pages, SLA reports, compliance evidence.

$199/month

Billed monthly, cancel any time, taxes may apply.

Start Business
What's included

10 status pages

25 agents

1,000 metric definitions

Unlimited SLOs

10 custom domains

150,000 email subscribers

180d metric history

Customer-scoped pages

SLA reports per customer

Executive dashboard

Compliance evidence pack

Outbound webhooks

Auto-incident creation

Custom CSS

Priority support

EnterpriseContract

Multi-product orgs, unlimited everything, priority support. Talk to us about a contract.

Contact us
What's included

Unlimited status pages

Unlimited agents

Unlimited metric definitions

Unlimited SLOs

Unlimited custom domains

500,000 email subscribers

Custom metric history

Customer-scoped pages

SLA reports per customer

Executive dashboard

Compliance evidence pack

Outbound webhooks

Auto-incident creation

Custom CSS

Priority support

SSO / SAML

Custom integrations

View the full comparison table and every feature included in each plan on the pricing page.

Frequently asked questions

We get a lot of questions. Here are the answers to the most common ones.

Categories

Still have questions?

If you can't find what you're looking for, we're here to help.

Get in touch