Say yes to AI. Keep sensitive data local.
Your people keep using AI. SendSafe swaps personal data for neutral labels on their laptop before the prompt leaves, and puts it back in the answer.
Available formacOSWindowsiOSAndroid
- Reported per team, never per person
- Works with the major AI tools
Film
See SendSafe at work in 1:43.
One Friday afternoon, one prompt, and what happens to the details in it.
Read the transcript
It's Friday, four fifty-five, and Sam from support has one ticket left. So he asks ChatGPT to write the reply. Emma's name, her email, her phone number. Straight into the prompt.
Here's the thing. That prompt is regulated. Emma's details fall under GDPR. Fines go up to four percent of global revenue. And under the EU AI Act, every company that uses AI has to help its people use it responsibly. Healthcare, government, automotive, legal, finance, the boardroom. Every team has an Emma.
Meet SendSafe. It runs on every work laptop, between your people and any AI tool. When Sam hits send, SendSafe swaps the sensitive details for placeholders. Emma becomes Person one. The same goes for the file he attaches. Emma's details come out of the Word document too. The AI still gets the whole question. Just not Emma's details. When the answer comes back, SendSafe puts the real details back in. Right on the laptop.
Why per team? Because context matters. Support needs the order details, engineering needs the code. The AI gets that. Just not the sensitive parts. Your IT team sets the rules per team, once. It's always on, and it works with ChatGPT, Copilot, Claude and more. The dashboard shows what was protected. Never what was said.
Four fifty-six. Reply sent. Emma's details never left the laptop. SendSafe. Say yes to AI. Keep sensitive data local.
How big is your AI risk? 6 questions, 2 minutes.
It runs in your browser. Nothing leaves the page until you ask for the report yourself.
You get an honest risk profile, the data that matters in your sector, and which team to start with.
What's your role?
Teams
Every team uses AI. And every team has data that shouldn't leave.
The work has to get done and AI saves time. So a customer email goes in, a staff list, a contract. The moment you send, SendSafe replaces the personal data it recognises with neutral labels, on the device. The AI still does its job.
See what leaves without SendSafe-
ChatGPTReplying to a customer about a refund.
Leaves the company for ChatGPT: the customer's name, email address and phone number.
-
ClaudeExplaining a payroll correction to a client.
Leaves the firm for Claude: the employee's name and IBAN.
-
MistralTightening a clause in an NDA.
Leaves the firm for Mistral: the names of both parties.
-
GeminiTidying up an appointment letter to a patient.
Leaves the hospital for Gemini: the patient's name, citizen service number and phone number.
-
Claude
Shortening a memo on a takeover.
Leaves the building for Claude: the names, email addresses and phone numbers of the directors involved.
-
CopilotDebugging a failed customer export.
Leaves the company for Copilot: names and email addresses from the log lines.
-
Copilot
Rewriting a permit letter in plain language.
Leaves the town hall for Copilot: the resident's name and citizen service number.
Illustration: Sam asks ChatGPT for a short refund confirmation, with the customer’s name, IBAN and email address in the prompt. Nothing happens while Sam types. On send, SendSafe replaces the three values with neutral labels on the laptop, so ChatGPT receives PERSON_1, ACCOUNT_1 and EMAIL_1. ChatGPT answers with those labels, and SendSafe puts the real values back in the answer on the laptop. The ChatGPT window always shows the real values.
Sam writes in ChatGPT with Daan’s name, IBAN and email address in the prompt. Nothing happens while Sam types.
On send, SendSafe swaps the details on the laptop. ChatGPT receives PERSON_1, not Daan.
The answer comes back with neutral labels. On the laptop, all three are back in it.
AI tools
Built to sit in front of the AI tools teams use today. On desktop and mobile, in the browser, the IDE and the terminal.
For IT
A control IT can sign off. The three things a security team asks first.
- Policy
One policy per team
Support keeps contact details local, Finance IBANs, HR names and citizen service numbers.
- Storage
The mapping stays on the device
Originals and the mapping stay where they were typed, so there is no new archive of employee prompts to guard.
- Reporting
Reported per team, never per person
The dashboard shows counts by team, category and AI tool. Never what was said, and never who said it.
Personal details fall under GDPR, and the EU AI Act asks every company that uses AI to help its people use it responsibly. SendSafe is a technical measure for one part of that: keeping personal data out of prompts.
SendSafe in short.
The questions teams ask first. More on How it works and Pricing.
What is SendSafe?
SendSafe is an app on work laptops and phones for companies whose teams use AI tools such as ChatGPT, Copilot and Claude: in the browser, in desktop apps, in the IDE and in the terminal. When someone sends a prompt, it replaces names, BSNs, IBANs, email addresses, phone numbers and cities with neutral labels and, on the laptop, puts the real details back in the answer. SendSafe is made by OneMinded.
Which AI tools does it work with?
It is built to sit in front of the AI tools teams use today, such as ChatGPT, Claude, Gemini, Copilot, Mistral and Perplexity, in the browser, desktop apps, the IDE and the terminal, and on mobile. Your policy stays the same when your team switches between them.
Does SendSafe read along while I type?
No. It acts at one moment: send. Nothing is checked while you type. It also does not replace your AI tools and does not build a prompt archive.
What stays on the laptop?
The original values, the mapping between values and labels, and the restored answer. Only the question and its context are sent, with neutral labels such as PERSON_1 in place of the details.
Why does this matter under GDPR and the EU AI Act?
Names, email addresses and phone numbers in a prompt are personal data under GDPR, and fines go up to €20 million or 4% of worldwide annual turnover. Article 4 of the EU AI Act asks organisations to help their people use AI responsibly. SendSafe reduces how much personal data reaches AI providers. It is a technical measure, not a full compliance programme. Read more. This is not legal advice.
Can we use it yet?
Yes, in a paid pilot. We start with one team, with criteria agreed up front. If it works, we expand.
Start with one team.
Pick the workflow where AI already helps most. We will set up the policy with you.