What each capability does and what it costs.

English, Arabic, French and Spanish at launch. You pick the default your team writes in, visitors switch with the toggle, and the multilingual feature publishes your incident in every language your audience reads. More languages are added to the catalog over time.

Arabic pages use a right-to-left layout, Arabic-Indic numerals, and dates and times formatted for Arabic readers.

Yes. Visitors subscribe with their email from the page itself and are notified when something goes down and when it recovers. It is double opt-in, so nobody is subscribed without confirming, and every message carries a one-click unsubscribe. The same updates can post to a Discord, Microsoft Teams or Telegram channel: you paste in a webhook from your own server or workflow, or your own Telegram bot, and no app has to be installed on either side. There is also an RSS feed and a status badge you can embed.

Sometimes you shouldn't. Self-hosted tools are good and free to license, but they are not free to run. They need infrastructure, plus setup, upgrades and backups, and a status page deployed beside your main systems can be taken out by the same outage. Multilingual incident publishing, including proper Arabic RTL, may also need additional setup. If you only ever need internal uptime alerts, self-hosting is a reasonable answer.

A website checked over HTTPS, a TCP port, or a DNS record: an API health endpoint, a database port, a domain that has to keep resolving. Website monitors can also check the response body for text or a JSON value; it's read to run that check and never stored. Heartbeat monitors cover cron jobs and pipelines: your job checks in, and a missed check-in raises the signal.

On managed infrastructure we run ourselves, kept separate from the systems it monitors. We store very little: check results keep status codes and response times only. A website monitor's response body is read only to check the assertion you configured for it, then discarded, never stored. AI translation sends nothing but the incident text you wrote to a third-party model provider, under terms that forbid training on it.

It is off until you turn it on, per monitor. When it is on, four rules bound it: nothing is published until a service has been down continuously for 15 minutes, never during a maintenance window, never on top of an incident you have already written, and never when our own checks look like the thing that broke. The text is written and reviewed by us in each language ahead of time (no model writes it while your site is down), and it names no cause and promises no time, because a check knows only that your endpoint stopped answering. The moment you edit it, it is yours and we stop touching it. Your team is alerted either way, whether or not anything is published.

After the paid period ends, Free keeps one public page on your InBrief subdomain and three HTTP, TCP or DNS monitors checking every five minutes, with alerts in one Slack channel. Additional monitors, pages and paid features stop running; their settings are retained for an upgrade. If the first charge leaves you unhappy, we refund it in full within 14 days.

Not without warning. An increase that reaches an existing subscription is announced by email at least 90 days before it takes effect, and applies from your first renewal after that date. A month you have already paid for is never repriced, and cancelling before that renewal means the new price is never charged.

No. New accounts get a 14-day trial of released Cloud features without entering payment details. Features awaiting release are not included.