SAFE MAILby Gatewatcher · for Microsoft 365
Home/How it works

An AI that reads every email like an investigator.

From delivered email to warned employee: what SAFE MAIL looks at, how it decides, and what stays in your hands.

How it works

From delivered email to warned employee: six steps, four seconds.

  1. 01The email is deliveredIn the mailbox, with no delay.
  2. 02Full readBody, links, attachments, QR codes.
  3. 03The AI judgesA danger score from 0 to 4, with reasons.
  4. 04Your contextAuthenticated colleague, trusted domains: your rules.
  5. 05The actionNothing, a label, or set aside. Never deleted.
  6. 06Your teamConfirms or returns the message, in every mailbox at once.

What the AI examines in every email

Who is writing?
  • Address one letter off
  • First message ever
  • Brand ≠ domain
What do they ask for?
  • Payment, new bank details
  • Password, sign-in
  • Urgency, deadline
Where do links lead?
  • Not the sender’s domain
  • Link hidden in an attachment
One emailread like an investigator would
What do attachments hide?
  • PDF, Office, images (OCR)
  • QR codes, attached emails
  • Disguised program
Does it add up?
  • Tone fits the sender
  • The request makes sense
And in your company?
  • Colleague authenticated or spoofed
  • Trusted domains
  • Existing conversation

One score, two thresholds, three outcomes

0.35Legitimate report, with attachment
3.99Fake delivery notice
023.54
Nothing to reportThe email stays where it is. Nobody is disturbed.
To verifyA clear label and one sentence saying why.
Set asideDedicated folder, one click to recover.

Scores measured on our test environment. Default thresholds (2 and 3.5), adjustable. At first, SAFE MAIL watches and touches nothing.

Protected before you finish your coffee.

1You sign inWith your Microsoft 365 account. No form, no settings, no sales call.
2The AI re-reads your weekIn two minutes you see the attacks that reached your people over the last seven days. Yours — not a demo’s.
3Your people are protectedEvery new message is read, for everyone, around the clock. There is nothing left for you to do.

The same fraudulent email in 12 mailboxes? Removed everywhere, in one click.

An attack rarely targets one person. As soon as our AI spots a fraudulent email, it finds everyone who received it. You remove it from every mailbox in a single click — even from those who already opened it. Change your mind, and you put it back.

Your spam filter cannot do this: once an email is delivered, it can no longer take it back.

Fraudulent emailInvoice pending — new bank details Received by 12 employees · 7 already opened it
CAClaire AubryOpened
MBMalik BessonUnread
LFLéa FontaineOpened
HMHugo MartinOpened
IRInès RocheUnread
PGPaul GirardOpened
SCSarah CohenOpened
TLTom LeroyUnread
JMJulie MorelOpened
KBKarim BenaliUnread
ADAnne DupuisOpened
NLNoé LambertUnread

An AI under your control.

It watches before it actsAt first, SAFE MAIL shows you what it sees and touches nothing. You let it act when you trust it.
It never deletes anythingA message set aside is one click away. No decision is irreversible.
It does not learn from your emailsMessage content is neither stored nor used to train a model.
It adds, and replaces nothingYour email and your current protection stay exactly as they are.
It stops in one clickNo notice period, and no trace left in your mailboxes.
It has a French publisherGatewatcher, a specialist in cyber threat detection.

Put it to the test. On your own email.

In two minutes our AI re-reads your week and shows you the scams nobody had seen. 7 days free.

Start free

Not the administrator? Send the invitation to the right person