Check-in

A heartbeat for a person

You promise to check in by a time; one tap keeps the promise. If the tap never comes, AlertRoster raises an incident to the people you chose — and keeps paging them until somebody acknowledges.

AlertRoster is a call-out notification and escalation tool — not an emergency service, and it does not contact one on your behalf. A check-in should be one of your arrangements, never the only one.

Silence is the event

Everything in AlertRoster is built on one spine: a source raises an incident, the incident is handed to whoever the source's escalation schedule says is on call, and the machinery pages, escalates, and records until somebody acknowledges. Monitors and infrastructure heartbeats have used that spine from the start.

Check-in adds a person to the list of things the spine can watch — with the logic inverted. A monitor raises an incident when an event arrives. A check-in raises one when an event doesn't: the absence of your tap is the alert. That inversion is the entire feature; everything downstream — delivery, escalation, acknowledgement, the timeline — is the same proven machinery every other alert already rides.

The life of one deadline

Arming a check-in starts two clocks: a reminder ahead of the deadline, and the miss at it. Which one matters depends only on whether you tap.

One armed deadline, three outcomes. The reminder is what makes the miss rare: it arrives while there is still time to tap or extend, and both actions are one press on the notification itself.

Two kinds, one state machine

A check-in is a persistent thing you own and re-arm — not a row per day. It is either idle or armed with a deadline, and every action is a named transition between the two.

The whole surface is five verbs — arm, check in, extend, cancel, and the miss the server applies. A daily never truly rests: answering tonight's deadline is what arms tomorrow's.

Timer — “expect me by 6:15”

Armed with an absolute deadline; rests once you check in, cancel, or it fires. Made for the solo hike, the evening run, the late drive home.

Daily — “every evening at 9”

A wall-clock time in your own time zone, rolling to tomorrow's occurrence each time it is answered — clock changes handled correctly. Made for living alone, working remotely, anyone whose quiet should be noticed.

Who gets alerted, and how

Every check-in points at an escalation schedule — a roster of the people who agreed to be reachable for you. Restaff the roster, add a second responder, or hand weekends to someone else without ever touching the check-in. If no schedule is set, a miss pages everyone in the account — loud on purpose.

A check-in is just another source. Everything to the right of “raises” is shared with every monitor and heartbeat in the product — the same delivery, the same escalation ladder, the same acknowledge that stands everyone down, the same recorded timeline.

Designed against false call-outs

The failure mode that ruins a tool like this is not the missed emergency — it is the forgotten timer that wakes a family at 3 am. Several deliberate choices guard against it:

The reminder is load-bearing. Ahead of every deadline — 15 minutes by default, tunable, or off — your phone gets a time-sensitive nudge with check in and extend right on it. Acting takes one press.
Extending is race-proof. The notification carries the deadline it was sent about; a stale tap is refused with the current one rather than silently extending the wrong clock.
A late check-in doesn't cancel the call-out. Once a miss has paged the roster, standing responders down is a decision made on the incident itself — someone may already be moving. The timeline keeps the honest record either way.
The clock has a backstop. Every armed deadline is watched by a sweep that re-arms any timer the system lost — a person's deadline is not silently dropped.

What a day with it looks like

The solo hike

Arm a timer for “back by 6:15” at the trailhead. Running long, press +15 from the reminder. Home safe, one tap and it rests. Forget entirely, and the two people you chose get a real page with your name on it — not a text that can sit unread.

Living alone

One daily check-in at 9 pm, roster: your two adult kids. Most nights it is a single tap at the 8:45 nudge. The night the tap never comes, both phones ring at 9:00 and keep escalating until one of them acknowledges — and the record shows who answered, and when.

Working alone

The same machinery reads as a working-alone procedure: an employee on a recurring check-in, the escalation schedule being the supervisor rotation, and the incident timeline being the written record that the procedure ran.

What it will never do is part of what it is

Check-in is heartbeat monitoring and absence detection for a person — and the things it refuses to become are commitments you can rely on, not roadmap gaps:

No sensors watching you

No motion, fall, or inactivity detection — nothing infers your state from a device. The only signal is the one you deliberately send, which is also why it never fires because a phone sat still in a bag.

Your people, not a call center

No one at Cloud Bedrock watches or responds to alerts, and the product never contacts emergency services on anyone's behalf. A missed check-in reaches the roster you chose — people who agreed in advance and answer by name.

AlertRoster is a call-out notification and escalation tool. It is not an emergency service and does not contact one on your behalf. It is not a fire alarm, a security alarm, or an alarm monitoring service, and it holds no life-safety certification. It should not be the only way a call-out can reach your people.