AI NetRiva
Self-hosted network monitoring

Quiet until something is wrong.

AI NetRiva keeps an eye on every device on your network and tells you the moment one stops answering — with enough context to fix it, not just a red dot.

14 days, everything switched on · Windows & Linux · amd64 & arm64 · one file, no database

RIGHT NOW LIVE
26UP
2LOSSY
1DOWN
NVR — dock camtimeout
AP — warehouse38.1 ms · 12%
Core switch1.2 ms
ASSISTANTThe dock camera has bounced 6× in 20 minutes while PoE draw sits at 94% of budget — suspect the line card.

Everything, on one machine

No agents to roll out, no collector to host, no database to install. Copy one file onto the machine that's always on and you're monitoring in about a minute.

Automatic discovery

Sweeps your subnets and fingerprints what it finds — switches, APs, cameras, printers, servers — using ARP, reverse DNS and MAC vendor lookup.

Continuous checks

Latency, packet loss, SNMP interface counters, CPU, memory and disk — sampled every few seconds, with rolling history drawn as sparklines.

Talk to your network

Ask what's wrong in plain language and it reasons over your own counters, logs and console output to name a cause. It can go further and draft the fix — a config change you read and approve before anything is sent to a device. Set recurring jobs to check things overnight on a token budget you cap. Your API key, your choice of provider.

Alerts you'll notice

A siren in the browser, email, ntfy push, or a webhook into whatever you already use. Acknowledge an alarm so it stops nagging, or let it escalate.

A map of the actual place

Drop devices onto your own floor plan or site photo, draw the links, and see what an outage takes with it. Neighbours come from CDP and LLDP.

Your network stays yours

Every measurement lives on your machine. Device names, addresses and configs are never uploaded — not to us, not to anyone. Exactly what does leave.

See what the switch itself says

Persistent SSH sessions stream console output off your switches into a live, searchable feed — the messages that normally scroll past on a cable nobody has plugged in. A port flapping, an STP change, a PSU complaining: you read it as it happens, next to the graphs, instead of finding it days later.

Runs without anyone logged in

Installs as a Windows service or a systemd unit, so it keeps watching when the machine reboots at 3am and nobody is sitting in front of it. A monitor that only runs while you are logged in is not a monitor.

Move it to another machine in minutes

Your whole setup — devices, thresholds, alerts, map layout, floor plan — downloads as a plain zip you can open and read. Restore it somewhere else and you are back where you were, without rebuilding anything by hand.

Built during shows, not in a meeting

It started as two Python scripts holding a network together while people depended on it. This is the rewrite — same job, built properly. The parts that exist are the parts that got used at 2am.

Setup guides that already know your address

The built-in knowledge base prints the exact SNMP or syslog command for your switch with this server's real IP and port already substituted in — so you copy a working line, not a template.

Syslog and live consoles

An RFC3164/5424 syslog receiver with search and retention, plus persistent SSH sessions streaming switch console output into the same feed.

Config backup with diffs

Pulls running-configs over SSH on a schedule, keeps versions, and shows you what changed between any two — which is usually the answer.

Try it on your own network

Free for 14 days with every feature switched on, up to 10 devices. After that, an annual licence per install — from $49/year. Updates and support are included while it's live.

Download See pricing

Windows & Linux · amd64 & arm64 · changelog

Questions before you buy?

Licensing, deployment, or a device it fingerprinted wrong. A real person replies within a working day.

The form reaches a person, not a ticket queue.