Watch out for them, without watching over them

NoordGuard sits quietly on the computer your child — or your parent — uses. A local AI looks at what is on the screen, stops the genuinely dangerous, then tells you what happened in plain words, and what to say next.

Analysis runs on the device · nothing is uploaded unless something is flagged · a Linux build is available too

The NoordGuard guardian window: three things needing review, and a list of incidents already handled
Plain language, not alerts

It says what happened — and what it already did about it

Every flag arrives as a sentence rather than a severity score: who it involves, what was on the screen, and where it came from. Underneath is a timeline of everything NoordGuard did before you opened it.

  • Serious, Important, Worth a look — words, not a colour-coded scale
  • The saved snapshot is blurred by default, so you can judge it without reading their messages
  • Serious flags are re-checked by a second, independent AI before they reach you
An incident written in plain language, with a blurred snapshot and a timeline of what NoordGuard already did
Then the harder part

A way to start the conversation, not just a block

Blocking a site is the easy half. Every serious incident comes with a few opening lines, what tends to help, and what tends to backfire — written for the thing that actually happened, not a generic safety leaflet.

  • Two or three ways to start, in the words a parent would use
  • What to avoid — taking the laptop away usually just teaches them to hide things
  • When something is worth reporting, it says so, and what to save first
Guardian actions and a conversation guide: ways to start, what to do, and what to avoid
Nothing to keep up with

Three things need you. Not three hundred.

Almost everything NoordGuard sees is nothing at all, and it stays that way. What is left is a short list: the ones waiting on a decision at the top, everything already dealt with underneath, each with a line saying how it was resolved.

  • One decision per item — keep the block on, or call it a false alarm
  • An email reaches you when something is serious; the rest simply waits
  • Handled items keep their history, so you can look back weeks later
The NoordGuard inbox: three incidents needing a decision, and three already handled
Your call, not ours

You decide how much of their screen you ever see

Watching over someone is a serious thing to do, so NoordGuard is built to take the least you need. Choose what a flagged event is allowed to show you — once — and it applies to every event after it.

  • Text only, blurred, or the full picture — blurred is the default
  • Snapshots are deleted from the device on a timer you set; 24 hours to begin with
  • Delete every event and every snapshot whenever you want, in one click
Privacy settings: choose text-only, redacted or full screenshots, and how long they are kept
Also in the app

Built for the rest of the household too

Not only for children

Set it up for a parent instead. The same detector catches the fake bank logins, prize pages and urgent-payment scams aimed at older people.

It knows their age

Give a birth year and what counts as serious moves with it. An eight-year-old and a fifteen-year-old don’t get the same threshold.

Works without the cloud

Choose local-only during setup and nothing is ever sent anywhere. Flagging, blocking and notifications all carry on as before.

Offline is fine

No connection? Events queue on the device and catch up when it comes back. Blocking never depended on the network to begin with.

Pause for an hour

Some moments aren’t for monitoring. Pause it from the menu bar and it starts itself again afterwards.

Your list, not just ours

Unblock something it got wrong and the site works again straight away. Add domains of your own and they stay blocked.

Nothing here is hidden

It doesn’t pretend not to be there

NoordGuard stays in the menu bar the whole time it is running, and the person at the computer is told when something is flagged. Blocked pages show a plain “site unavailable” — no red warning, nothing to be ashamed of. And the app explains itself, in the same plain words, to whoever opens it.

  • What it looks at, what leaves the device, and how long anything is kept — written out in the app
  • Analysis happens on the machine itself, on a model that ships inside the app
  • Nothing is uploaded unless something is flagged, and then only what you allowed
The in-app explanation: what we look at, what leaves the device, what the monitored user sees, and how long things are kept
One quiet install

The conversation goes better when you already know

Put it on the computer they use. Setup takes a few minutes, and after that it stays out of the way.

Screenshots are from the current build, filled with sample data — the names, sites and incidents shown are invented.