Skip to content

Your data under control

Your data works,it stays yours.

A report, a connection, an assistant or the AI medical secretary reads your data. Here is what it reads, what it never touches, who has access, where the data goes when an AI reads it, and how to cut everything off.

Read-only

What we read, what we never touch

When a report or an assistant needs to read your data, we open an access that only allows reading. It isn't a promise on our part, it's a setting: even if someone asked it to change an invoice, it couldn't.

At one clinic, we set this lock twice. The connector refuses any write request before it even leaves, and the accounting account itself only has read rights. We tested write attempts before handover: they were all blocked. Writing is only opened if you decide so, for a specific task, and never for a payment.

The rules

Four rules, on every connection

They apply to a report, a connection between programs, an assistant or the AI medical secretary.

01

Who has access

One account per person, never a shared password. Everyone only sees what they already see in their software.

02

Where your data goes when an AI reads it

It stays in your software and is read at the moment of the question. A written framework states which AI service is used, where it is hosted and how long it keeps anything.

03

Patient data

Health data, medical confidentiality, staff: nothing is connected without a written framework agreed with you before the first line. Often the right answer is not to connect it.

04

Cut everything off, at any time

Each connection has its own access. You withdraw it whenever you like, without calling us, and the connection stops without writing anything wrong.

What we don't do

The machine prepares, a person decides

We don't automate a decision. A payment reminder can be prepared automatically, but a person sends it. A message to a patient can be drafted, but a person approves it. When a mistake would cost more than the time saved, the machine prepares and a person decides.

  • Never a payment triggered by a machine.
  • Never an AI pretending to be a person, and never a diagnosis made by an AI.
  • Never a robot clicking in a program in your place.
  • Never patient data connected without a written framework.

Your questions about your data

How do we keep control over who can access our data?

One account per person, read-only by default, and a dedicated access per connection, which you can withdraw at any time. Everyone only sees what they already see in their software.

Can we use ChatGPT, Claude or Gemini on our emails and our accounts, with human oversight?

Yes, if the framework is set first. The AI is read-only, everyone signs in with their own account, and a person decides before any action: nothing is written or sent without them. That is how we set up the assistant connected to a clinic's accounts.

Which AI reads our data, and where is it hosted?

The written framework states it before opening: which service, hosted where, and what it keeps. You approve it. The connector that reads your software is hosted by DeepIT.

Can we use ChatGPT or another AI with patient data (GDPR)?

We don't connect it without a GDPR framework agreed with you: which data, for what use, hosted where, and what never leaves. Often the right answer is not to connect it at all.

A question about your data?

Tell us what you want to connect, and what worries you. We'll tell you what is possible, and what we advise against.