Skip to content
IRIS

The lifecycle

From signal to proof.

Every critical alert follows one stateful flow — not a fragile chain of automations. Here is exactly what happens between a sensor firing and an auditor signing off.

Start free — 1,000 credits View the API
  1. 01
    Signal

    A signal arrives

    Any sensor, monitoring system or script posts an alert to one REST endpoint. A freezer drifts +2.1 °C over limit, a line vibrates toward failure, a substation faults at 3 AM — IRIS treats every critical signal the same way.

    • One idempotent POST /api/v1/alerts call
    • Severity, group and per-language message in the payload
    • No SDK lock-in — works from any device or platform
  2. 02
    Dispatch

    Dispatch on the right channels, in the right language

    IRIS resolves the on-call group, picks each recipient’s language and the best sender number, and sends across the live channels — SMS and Voice/IVR today — in parallel. The alert becomes a stateful object the moment it lands.

    • Per-recipient language (6 supported) and number mapping
    • SMS + Voice live; Slack, Teams, WhatsApp, email on the roadmap
    • Each send tracked for delivery status and actual cost
  3. 03
    Escalate

    Escalate round by round until acknowledged

    No acknowledgement in the configured window? IRIS escalates to the next person in the on-call chain — up to ~20 rounds — without dropping the thread. The state machine remembers exactly who was tried, when, and on which channel.

    • Configurable wait windows between rounds
    • Up to ~20 escalation rounds per alert
    • Survives restarts — the alert state is durable, not in-memory
  4. 04
    Acknowledge

    A human acknowledges

    The recipient replies “1” to acknowledge (or “2” to escalate) by SMS, or presses a key on the voice call. IRIS interprets the reply, stops the escalation, and records who acted and when — closing the loop two ways, not one.

    • Reply “1” to acknowledge, “2” to escalate
    • Keypad acknowledgement on voice calls
    • STOP/START opt-out handled in 6 languages
  5. 05
    Prove

    Prove it with an audit trail

    On acknowledgement the alert resolves and IRIS writes an audit-grade record of the whole event — every send, escalation, reply and timestamp — with configurable retention. Exportable for GxP, HACCP and ISO audits.

    • Immutable event log: reached, acknowledged, proven
    • Configurable retention; GDPR Art. 15 export & Art. 17 erasure
    • EU data residency by architecture on Cloudflare’s edge

Why stateful

Not a Make/Zapier chain

A no-code chain fires a step and forgets it. There is no memory of who was tried, no timer that survives a restart, no way to resume an escalation mid-flight — and no record an auditor will accept. We know: AMS v1 ran on 26 Make.com scenarios before we rebuilt it.

IRIS models each alert as a durable state machine on Cloudflare Durable Objects. The alert is the state — it remembers, waits, escalates, and proves resolution, even across deploys and failures.

The escalation engine
  • Durable timers survive restarts
  • Resumes escalation mid-flight
  • Remembers every attempt & reply
  • Audit-grade event log, by default
  • EU data residency by architecture
  • One endpoint, no brittle chain

See it run on your alerts

Spin up a trial workspace and send your first escalating alert today — 1,000 credits, no credit card.

Start free — 1,000 credits See pricing