Two typical scenarios: Notifly notices an outage on its own and makes sure you hear about it — and an AI agent asks for your permission before it acts.
A monitor checks your site from the cloud — say, once a minute — and sees that it isn’t responding.
A push arrives on your phone and in the web dashboard.
5 minutes pass and nobody has read the notification.
Escalation: Notifly sends an SMS, calls or writes to Telegram — whatever you set up for the channel.
“Read” — and the escalation stops.
The agent calls /ask (via MCP, an SDK or curl) before a risky
step: “Roll out v2.4 to prod?”
A notification with buttons “Yes” and “No” arrives on your phone.
You tap a button right in the notification shade — or answer in the web dashboard.
The agent gets the answer and carries on. No answer in time — the default answer kicks in.
Sign up with your email — free, no card required — and confirm your address. First login →
The first-run wizard creates a channel and shows its token together with a ready-made send command.
Install the Android app and sign in with the same account, or keep the web dashboard open. Send a test message — it should appear on the device.
curl -X POST "https://api.notifly.ru/message" \ -H "X-Notifly-Key: <APP_TOKEN>" \ -F "title=Deploy finished" \ -F "message=v2.4.1 rolled out to prod in 38 seconds" \ -F "priority=5"Replace <APP_TOKEN> with the channel token — the wizard shows this command with your token
already filled in. No workers, queues, or your own backend required.
REST API & WebSocket
Send messages with a single HTTP request from any language, script or CI/CD, and receive the stream of new messages over WebSocket. → Sending messages · → WebSocket
Delivery & escalation
Push not read within N minutes — the message goes out by email, SMS, a phone call, to Telegram, Slack, Discord, Teams, Mattermost, PagerDuty or Opsgenie. Marking it read stops the escalation. → Delivery
Active monitors
Cloud probes: HTTP, page content, ports and protocols, SSL certificates, workflows and browser scenarios with screenshots — with alerts on failures and recoveries. → Monitors
Heartbeat
Dead-man-switch: ping Notifly every minute, hour or day and find out when a task has stopped arriving. → Heartbeat
Alertmanager, Grafana & Zabbix
Native alert receivers: a ready-made receiver for Prometheus Alertmanager and Grafana, a media type for Zabbix, severity → priority. → Monitoring integrations
Webhook router & Email Inbox
One URL for webhooks from GitHub, GitLab, Bitrix24 and payment providers — rules route the
events to channels. And emails sent to <alias>@in.notifly.ru become notifications.
→ Webhook router · → Email Inbox
Questions in channels
Ask a question with buttons, a free-text answer or a form — and get the decision straight from the notification. Timeout and default answer for approval gates. → Questions
MCP server
Notifly as Model Context Protocol — your AI agent reads and sends messages itself and calls you when a decision is needed. → MCP
Quiet hours & maintenance windows
At night only urgent messages wake you, the rest comes as a morning digest. During planned work monitors stay silent and uptime isn’t hurt. → Quiet hours · → Maintenance windows
Status page
A public page with 90-day uptime, an incident feed and email subscriptions for your customers — built from the channel’s monitors. → Status pages
Metrics
Send numeric metrics from apps and servers, build charts and alerts on thresholds. → Metrics
Web-script
One line of JS — and the site starts sending events and errors directly to you. → WebScript
Also: the AI assistant in the dashboard, the CLI, SDKs, channel sharing — all in the documentation.
Manage channels, devices, monitors, tokens and delivery. A message feed with search and a
priority filter, a Cmd/Ctrl+K command palette, light and dark themes. Sign-in is protected
with a second factor: a TOTP app, sign-in approval from your phone and recovery codes.
API message → push to your phone in fractions of a second. Act right from the notification: “Read” stops the escalation, option buttons answer a question, and a free-text reply is typed in the shade itself. The feed works offline, monitors have their own native screen, and quiet hours are set up in the app. The app is distributed as an APK file from this website.
Every feature on every plan: monitors, escalation, questions, MCP. Plans differ only in limits.
299 ₽ / month
100,000 events a day
15 devices
5 status pages, Notifly badge can be hidden
999 ₽ / month
Unlimited events
Unlimited devices
Unlimited status pages
What counts as an event. One action is one event: a message sent, a monitor check, a heartbeat ping. A monitor that checks every 5 minutes uses 288 events a day. The counter resets at 00:00 Moscow time.
How to pay. By bank transfer: write to support@notifly.ru — we’ll send you an invoice, and once the payment arrives we’ll switch your account to the plan. Card payment is not available yet.
All limits and details are on the “Quotas and plans” page.
Accounts and messages are stored in Yandex Cloud data centres in Russia; ainotifly.com only runs its public edge on AWS.
Telegram, Slack, PagerDuty and other services receive messages only if you connected them for delivery yourself.
Two-factor authentication: a TOTP app, sign-in approval from your phone and recovery codes.
Learn more: Privacy Policy · Account security
You can start for free: the Free plan is 0 ₽ with 1,000 events per day, and sign-up requires no card. Pro costs 299 ₽ per month, Business — 999 ₽ per month. Every feature is available on every plan; only the limits differ: events per day, devices, status pages, AI requests, port scans and screenshot storage. An event is a single action: a message sent, a monitor check, a heartbeat ping.
Learn more: Quotas and plans
By bank transfer against an invoice. Write to support@notifly.ru with the plan you need (Pro or Business), the period and your account email — we will reply with an invoice and bank details. Pay it, putting the invoice number and your account email in the payment reference. Once the payment arrives, the operator switches your account to the new plan and lets you know. Card payment is not available yet.
Learn more: Quotas and plans
Notifly is a single API and message history with priorities, monitoring and metrics. Push reaches the app and the browser instantly, and if a message isn't read, Notifly re-sends it by email, SMS, a phone call, to Telegram, Slack, Discord, Microsoft Teams, Mattermost, PagerDuty or Opsgenie. Telegram is just one delivery channel, not the whole system.
No. Notifly is a SaaS: no workers, no queues, no backend of your own. A single HTTP request (even curl) is enough to deliver a message.
Inbound: REST API, WebSocket, Email Inbox (an address like <alias>@in.notifly.ru), a Webhook router for GitHub, GitLab, Bitrix24 and any other webhooks, native receivers for Prometheus Alertmanager, Grafana and Zabbix, WebScript for websites, a CLI and an MCP server for AI agents. Outbound: email, SMS, phone call, Telegram, Slack, Discord, Microsoft Teams, Mattermost, PagerDuty, Opsgenie and a custom webhook.
In the Android app (push to the shade within a fraction of a second, with “Read” and answer buttons), in the web dashboard — while its browser tab is open — and in any MCP client. Anything left unread goes further along the escalation you configure — by email, SMS, a phone call or to team chats. The Android app is distributed as an APK from the website.
Learn more: Android app · Delivery & escalation
Not publicly yet: the iOS client is not published on the App Store or TestFlight — it can only be built from source, and it shows notifications only while the app is open. If you use an iPhone, set up escalation to SMS, a phone call or Telegram so important messages still reach you, and use the web dashboard in your browser.
Learn more: iOS application · Delivery & escalation
Check in order: is the device connected and active (the “Devices” section of the dashboard); is the channel muted on that device; are quiet hours or a maintenance window in effect right now; is the daily event quota used up. On Android the most common cause is battery optimisation: turn it off for Notifly so the system does not kill the app.
Learn more: Channels & devices · Quiet hours · Maintenance windows · Quotas and plans · Android app
That's what quiet hours are for: during the window only urgent messages (priority at or above the threshold) are escalated, and the rest arrives in the morning as a single digest. For planned work, create a maintenance window — monitors keep checking, but alerts and escalation are held back and the status page keeps its uptime.
Access to every channel is protected by a token, API tokens are limited by channels, tags, access level and expiry, and channels can be shared with limited permissions. Sign-in is protected with a second factor: a TOTP app, sign-in approval from your phone and recovery codes. Account data and messages are stored in Yandex Cloud data centres in Russia; ainotifly.com only runs its public edge (API, ingestion, WebSocket) on AWS. Messages go to external services (Telegram, Slack and others) only if you connect them yourself.
Learn more: Account security · Privacy Policy