by OneMinded
EN EnglishNL Nederlands
Book a demoDemo

How it works

How SendSafe works. One request, from keyboard to AI and back.

For teams that work with AI every day. SendSafe runs in the background on the device and swaps personal data for neutral labels as a request is sent. Follow one request here.

Book a demoWatch the video

Illustration: an employee's prompt to ChatGPT names a customer, her email address and phone number. As it is sent, SendSafe replaces the three values with neutral labels on the device. The answer comes back with the labels, and the originals are restored on the device.

One request, from start to finish.

A support reply drafted in ChatGPT, followed from the keyboard to the provider and back.

  1. A support employee writes as usual

    The prompt is about a customer, Anna de Vries. Her name, email address and phone number go in, the way they always have.

    On device

  2. The employee presses send

    This is the one moment SendSafe acts. The prompt is checked against the team's policy before anything leaves.

    On device

  3. Her details become neutral labels

    Anna de Vries becomes PERSON_1, her email EMAIL_1, her number PHONE_1.

    On device

  4. Only the version with neutral labels is sent

    ChatGPT gets the question and the context. Just not Anna's details.

    Sent

  5. The answer is restored locally

    The reply comes back with the same labels. On Mac and Windows, SendSafe puts the originals back on the device, so it is ready to use.

    On device
Typedon the device
Draft a short reply to Anna de Vries at anna@devries.example. Her refund is approved, and we will call her on +31 6 1420 2287 on Monday.
Sentleaves the device
Draft a short reply to PERSON_1 at EMAIL_1. Her refund is approved, and we will call her on PHONE_1 on Monday.
Returnedfrom the provider
Hi PERSON_1, good news. Your refund has been approved. We will call you on PHONE_1 on Monday to confirm.
Shownon the device
Hi Anna de Vries, good news. Your refund has been approved. We will call you on +31 6 1420 2287 on Monday to confirm.

What it deliberately leaves alone.

A protection layer earns trust by doing one thing well and staying out of everything else.

It does not watch typing

SendSafe has no live flagging, pop-ups mid-sentence or keystroke checking.

It does not replace your AI tools

People keep using the assistants your organisation has approved. SendSafe is a layer around that use, not another chat app.

It does not build a prompt archive

Originals and the mapping stay on the device. Employee prompts do not end up in a new central store.

What stays on the device, in detail

Rollout

Through what IT already uses. Set up once, then everyone keeps working as they do now.

  1. IT

    Rolled out by your IT lead

    Through the device management you already have.

  2. Rules

    Set once per team

    After that they stay in place.

  3. Employees

    Nothing changes

    They keep using their AI tools the way they do now.

Questions about how it works.

Rollout questions are answered on Pricing.

Which values does SendSafe replace?

EU direct identifiers: name, BSN, IBAN, email address, phone number, city. You set the policy per team. How team policies work

Why neutral labels instead of deleting the values?

Neutral labels such as PERSON_1 keep the sentence intact, so the AI can still write a useful, specific answer. Because the mapping is kept locally, the originals can be put back afterwards.

What happens with follow-up questions?

SendSafe keeps the label mapping on the device for the whole conversation. A follow-up that mentions the same person again reuses PERSON_1, so the AI can keep the thread. The AI's chat history only ever holds labels, and on Mac and Windows the real values are restored locally when an answer is shown. Your organisation's retention setting decides how long the mapping is kept.

See it in your own workflow.

We set it up around the AI tools and data your team works with.

Book a demoWatch the video
Available formacOSWindowsiOSAndroid