What shipped, and when.
A paid product should be able to show its work — especially one whose pitch is that nobody sunsets a business model. This is the subset you would notice, in the words you would use. Pricing changes are in here too, at the same size as the features.
-
Monitors admit when they have never been checked
A newly created monitor used to show as up before a single check had run — a green light that had measured nothing. It now reads pending until the first result arrives, all the way through to the public status badge. Alongside it: a health advisor that names the eleven ways a setup can be quietly broken (no alert channel, a monitor nobody owns, an interval that outruns the plan budget) and offers to fix each one, plus an honest per-channel status so a channel that has never delivered anything says so.
-
Status pages on your own domain, with the certificate issued for you
Point
status.yourcompany.comat us; we verify the DNS record and provision the certificate. Team includes one domain, Agency ten, and pay as you go adds them from the wallet at €2 a month each. Pages can also be restricted by IP allowlist or access link, so an internal page is protected rather than merely unlisted. -
A new mark
UP with a negative-space 4, across the app, the operator console and this site. Same product, fewer apologies about the old logo.
-
Checker capacity that can grow without us waking up
The worker fleet is now provisioned through an operator-authorised lifecycle across two providers, with capacity alerting and guarded removal. You will not see it, which is the point: it is the machinery that keeps a 10-second interval a 10-second interval when the fleet gets busy.
-
Grace mode: zero balance no longer means dark
Running out of credit used to pause monitoring. Now a 7-day grace period keeps every monitor running at a 5-minute cadence, free of charge, with alerts and status pages live — and the runway warnings that precede it are based on your real burn rate rather than a percentage threshold. Only after those seven days do checks pause, and nothing is ever deleted.
-
One rate for every check type
HTTP, keyword, body, TCP, DNS and ICMP ping checks all cost €9 per million. Previously the meter charged different rates by protocol, which meant the cheapest way to describe our pricing was a table. Now it is a sentence. Non-check meters — alerts, status pages, seats — are telemetry only and cost nothing.
-
The region picker only offers regions we actually run
Checkers now report their own presence, and the picker is built from that signal. It is no longer possible to configure — and be billed for — a region we do not currently serve.
-
Slow is degraded, not down
A response that arrives late is now its own state: recorded, visible on the chart as an amber band, and never a reason to page anyone at 4 a.m. Confirmation is symmetric — three consistent observations in either direction — so a server having a bad minute no longer produces an incident and an apology.
-
Live on upfour.io
Production moved to a three-server split topology in Germany: application, dedicated database and a separate checker box, with only port 443 exposed. The status page deliberately does not run on it.
-
Reports you can hand to a client without editing them
Five report types — executive summary, per-monitor SLA with business-hours masking, downtime ledger with a public verification hash, incident analytics and response time — exportable as CSV, JSON, XLSX or PDF, schedulable by email, and white-label with your logo and domain. Share links freeze a snapshot so the number in the email is still the number three months later.
-
Predictable retention instead of a silent trim
Individual check results live at most 7 days in the primary database, then as rollups and a compressed archive for your plan’s window: 30 days on pay as you go, 90 on Team, 365 on Agency. Reporting reads across both tiers, so a 90-day chart is a 90-day chart.
-
Ping monitoring that is actually ICMP
Ping checks send real ICMP echo requests rather than opening a TCP connection and calling it a ping. Also shipped: a safety layer with a resolve cache, an IANA-reserved-range denylist and per-organization rate limits, because a monitoring service that can be pointed at arbitrary hosts is an amplifier if you let it be.
-
Client accounts
Every monitor, incident, status page, alert plan, playbook and report can belong to a client, partitioned at the database level rather than filtered in the interface. With it: read-only share links that need no account, and client templates so onboarding the next client ends before the kickoff call does.
Older than this list? The product existed before it was worth writing down — status pages, incidents, escalation, on-call and heartbeats all shipped between February and April 2026.
Still shipping. <em>Come and watch.</em>
€3 credit · no card · every feature from day one