Your site is up.
Is checkout working?
PylonMon does what a customer does — opens your site, signs in, adds to cart, pays — from outside your network, every few minutes. When a step breaks you get paged with the step, the error and a screenshot. Then it keeps watching everything underneath: uptime, servers, cron jobs, error rates, logs.
goto https://shop.example.com click Sign in fill Email = probe@shop.example.com fill Password = •••••• # stored encrypted, never shown again click Add to cart click Checkout expect Order summary
Your website can return 200 and still be broken
An uptime check tells you the server answered. It cannot tell you that nobody has been able to pay since 2am.
- the website responded
- with a 200
- in 180 ms
- login is broken
- checkout doesn’t work
- a button stopped doing anything
- your signup flow is failing
- the workflow your business depends on is down
- which step failed, and why
- what the browser saw — a screenshot of the failing page
- how long every step took, run after run
- the moment it recovers
And it pages the person on call: Slack, email, SMS, a phone call, an escalation ladder if nobody answers.
The real thing, not a mock: step 3 failed, here is the error, here is what the browser saw. Re-checked immediately, then paged.
Not sure how to write yours? Send the flow to setup@pylonmon.com and the PylonMon team will build the first journey with you on a call. No charge.
Monitor the flows your business depends on
One journey per flow. If a customer can’t do it, you know before they tell you.
Can customers pay?
Add to cart, checkout, order confirmed — walked by a real browser on a schedule. The most expensive thing to find out from a customer.
Checkout monitoring →Can users sign in?
Sign in, land on the dashboard, see their name. The password is stored encrypted and never shown again.
Login monitoring →Can new users register?
Fill the form, submit, expect the welcome. The flow that fails quietly the day a form field is renamed.
Write a signup journey →Can they do the one thing?
Book, upload, search, submit a claim — whatever the business runs on. Up to 25 steps per journey.
Journey steps reference →And everything underneath it
Journeys tell you the experience works. The rest of PylonMon tells you why it stopped — from outside your network, so the alert still goes out when your environment goes dark.
Websites, APIs, ports
HTTP, TCP, ping, DNS and UDP every 30 seconds, from off-site. Paid plans confirm an outage from a second region before anyone is paged — a one-region routing blip never wakes you.
Inside the box
An open-source agent pushes CPU, memory, disk, temperature and logs. Outbound only, no open ports, works behind NAT. Silence is itself the alert.
The dead-man’s switch
One curl at the end of the job. Miss the window and PylonMon pages you — the backup that quietly stopped running is finally loud.
Before it’s an outage
Error-rate rules on HTTP monitors, log search and live tail, log signatures that group failures by shape and page when a new one appears.
The right person, once
Slack, Discord, Teams, email, SMS, voice, webhooks. Escalation ladders, on-call rotations, incident timelines and public status pages — included, not sold as three products.
We page your agent
An AI-agent alert channel and an MCP endpoint let your agent investigate with your keys and write findings back to the incident. We never hold your credentials.
Every monitor on one screen, worst first: journeys, servers, heartbeats and HTTP checks with 90-day uptime and latency at a glance.
What happens when something breaks
Confirm, page, hand off, write it up — on one timeline, with the evidence attached.
- Detected and re-checked. A failure is checked again immediately; paid plans also confirm it from a second region. One flaky probe never pages anyone.
- Paged with the evidence. The alert carries what we saw: the failing step and screenshot, the HTTP status and body, the certificate, the log lines around it.
- Escalated if nobody answers. Slack, then SMS, then a phone call — your ladder, your on-call rotation.
- Acknowledged and worked. Paging goes quiet, checks continue, notes and runbooks live on the incident.
- Resolved with a write-up. Time to acknowledge, time to recover, and what happened — ready for the morning.
Built to be trusted with production
Read how it’s built before you point it at anything that matters.
Simple pricing
Start free. Upgrade to Pro when you need more journeys, faster checks and a second region confirming before you’re paged. Flat monthly pricing, never per seat.
The price on this page is the price on your invoice. No per-host meters, no metered browser runs, no usage tier that spikes the month something goes wrong. Detection, alerting and incidents are one product here, not three.
- 1 customer journey — a real browser walks your checkout every 15 minutes
- 20 monitors
- Log search — 1-day retention, 50 MB/day
- 1 alert channel — email, Discord, Slack, webhook, ntfy or web push
- 1-minute check intervals
- Heartbeat, HTTP/S, TCP & ping monitors
- pylon-beacon — 3 nodes (CPU/mem/disk/temp, open source)
- Escalation ladders
- 24-hour event history
- 5 customer journeys, every 5 minutes
- 75 monitors
- 10 alert channels
- Multi-region confirmation — an outage is verified from a second region before we page you, so a one-region routing blip never wakes you
- Log search — 7-day retention, 1 GB/day
- 30-second check intervals
- pylon-beacon nodes up to your monitor cap — with vital threshold alerts
- Prometheus & Alertmanager — metric monitors (push or pull) + your alerts as monitors (up to 25 alert streams)
- SMS allowance each month
- + DNS, UDP, TLS-cert & domain expiry, slow-response monitors
- Incidents, public status page & API
- Custom webhook payloads & maintenance windows
- Scrape your monitors into your own Prometheus
- 30-day event history + CSV export
- Priority email support
- Capacity blocks — +25 monitors $5/mo · +5 journeys $10/mo, prepaid, added by you, never an overage
- 20 customer journeys, every 5 minutes
- 250 monitors
- 200 alert channels
- Everything in Pro
- On-call rotations — shifts, overrides, page-whoever-is-on
- Company-scale Prometheus — up to 200 inbound alert streams (Pro caps at 25)
- Bigger SMS allowance
- Telegram alerts · WhatsApp (planned)
- Teammates & roles (up to 5)
- 90-day event history
- Log search & live tail — 30-day retention, 5 GB/day
- Capacity blocks — +25 monitors $5/mo · +5 journeys $10/mo, prepaid, never an overage
For organizations with security, compliance, availability and support requirements beyond the standard plans: custom limits, an SLA, a security review with an engineer.
- 1000 monitors
- Unlimited alert channels
- Log search — 90-day retention, 20 GB/day (soft — talk to us past that)
- Everything in Team
- Bring-your-own-AI — alerts page your agent (MCP endpoint + agent-scoped keys); it writes findings back to the incident
- Voice-call alerts & on-call voice paging
- Unlimited teammates — invited or SSO-provisioned, no seat wall
- Enterprise SSO (OIDC)
- Largest SMS allowance
- 24/7 priority support — round-the-clock, ahead of every other queue
- 100 customer journeys; custom limits on request
Questions, answered
The things people ask before signing up.
What is journey monitoring?
A journey is a short script — click Sign in, fill Email, click Checkout, expect Order summary — that PylonMon runs in a real browser from outside your network every few minutes. If a step fails you are paged with the step, the error and a screenshot. It catches the failures an uptime check cannot see: a login that errors, a checkout button that does nothing, a signup form that stopped submitting.
Is there a free plan?
Yes — Free is free forever: one customer journey, 20 monitors, 1-minute checks, heartbeat, HTTP/S, TCP and ping monitors and one alert channel. No credit card required.
What is heartbeat (dead-man's-switch) monitoring?
Your cron jobs, backups, and scripts send a quick HTTP ping to a unique URL when they finish. If PylonMon doesn't hear from a job inside its expected window, it pages you — the silence is the signal, so you catch jobs that die quietly.
Do I need to install an agent?
No. PylonMon watches from off-site: journeys run in our browsers, heartbeats are a single curl line, and HTTP, TCP and ping checks run from our side. The open-source pylon-beacon agent is optional, for CPU, memory, disk and logs from inside a machine.
How is PylonMon different from UptimeRobot or Healthchecks.io?
It does both at once — outbound uptime checks and inbound heartbeat/cron monitoring — plus escalation ladders on every plan, multi-region outage confirmation on every paid plan, on-call rotations on Team, Prometheus and Alertmanager support, and alerts to email, Discord, Slack, SMS, WhatsApp and voice.
What alert channels are supported?
Email, Discord, Slack, Teams, Webex, Telegram, ntfy.sh, web push, SMS, WhatsApp, voice calls, Zapier and any webhook — with escalation on every plan and one alert per state change, no 2am nag loops.
Can it monitor cron jobs and backups?
Yes — heartbeat monitors are built for exactly that. Add one curl line to the end of your job; miss the schedule and you're alerted with how long it's been silent.