Raydo is built on the European rules for AI.
European law gives you rights over your own company data. The people who appear in that data get rights over what an AI does with it. Those rights only apply where that law applies. So Raydo has run in Europe from day one.
On 2 August 2026 the transparency rules of the AI Act came in. They ask for three things: you have to know when you are dealing with AI, you have to be able to see where an outcome came from, and a person decides what goes outside. Raydo did all three before it had to.
So you never have to wonder whether it will still be allowed a year from now.
The AI Act, in Dutch the AI-verordening, is the European law for artificial intelligence. It has applied since 1 August 2024 and comes into force in stages. In February 2025 the banned uses came into force, in August 2025 the rules for the large language models themselves.
On 2 August 2026 the transparency rules came into force, article 50. That is the part that is about Raydo. The rules for high-risk systems follow later, in December 2027. One transition period still applies: systems that were already on the market before 2 August 2026 have until 2 December 2026 to mark what they make, so that a computer can recognise it.
Raydo came to market after that date and does not use that transition period.
Read the regulation at the European CommissionThat is how it already works, with European AI and European cloud. With that, Raydo keeps to the European AI Act and to the GDPR, and not to American law.
Your data sits with Scaleway in Amsterdam.
The language models run at Mistral in Paris, or self-hosted in Amsterdam.
Storage and language models both sit in Europe, and we keep to the European rules that go with that.
You use WhatsApp, Teams and Gmail to share messages. For as long as a message sits there, that party's terms apply. So we keep the work itself. In the channel you only get the notice, and you open the work with a secure link. You switch a channel on yourself, and you see exactly where that line sits while you do.
We never send passwords, keys or full exports through a channel. Under no setting.
Your data and everything Raydo learns about your company are yours.
A digital colleague that gets to know your way of working gets smarter from it. That knowledge stays with you, and never ends up with another company.
A digital colleague is only as reliable as the places the information comes from. From these three, Raydo builds your customer pictures, your meeting briefings and your follow-up emails. Nothing comes in that does not come from here.
Your CRM, your calendar, your mailbox. Only what you connect, and only for as long as you leave it connected.
Sources such as the Chamber of Commerce register, and databases we hold a licence for. We fetch that data straight from the source.
What you tell Raydo, what you upload and what you record.
Raydo does not comb through websites and does not empty out social networks. Raydo builds no profile from data someone never shared for this purpose. That is not a setting anyone can switch on.
If something is in a message that we did not get from the recipient themselves, we say so in that same message. We say what kind of data it is and where we get it from.
Raydo gets the work ready and leaves it standing. You press Send, and that is how it is built.
Raydo says nothing about a customer without showing where that came from.
Everything that goes outside carries that note, in the recipient's language.
Raydo makes no profiles of your customer's employees.
Those rights are in the GDPR, the European privacy law that is called the AVG in Dutch. Your customers have them over their own data, including when it sits with you. Raydo is built so that you meet those rights without making a project of it.
If someone wants to know which data you have recorded about them, you fetch that in one action. You can have data corrected or deleted. With every piece of data it says where it came from and when Raydo fetched it, so you can back up your answer.
If a check turns up something odd or suspicious, from a public register for instance, Raydo never decides on that itself. Raydo always validates it with you before it takes action, with the source alongside.
With Scaleway in Amsterdam. The language models run at Mistral in Paris, or self-hosted in Amsterdam. With that, everything stays inside Europe.
We destroy your key, and with that everything is unreadable. Until that moment you can fetch everything and take it with you.
No. What Raydo learns about your way of working stays yours and never ends up with another company.
No. Raydo gets the email ready and leaves it standing. You read it over, change it if you want to, and press Send yourself.
Yes. Raydo introduces itself as AI on the first message, and everything that goes outside carries that note in the recipient's language.
There is no back way in for ourselves. If someone of ours has to look inside your environment, during an outage for instance, that is a separate action. It goes into a log you can inspect. Nothing can disappear from or change in that log afterwards.
No. Raydo works with what you connect, with official registers and licensed databases, and with what you provide yourself. Scraping is not part of that.
Yes. Raydo is built to the GDPR, called the AVG in Dutch. Your data sits in Europe, with every piece of data it says where it came from and when it was fetched, and you meet the rights of the people in your data in one action. What Raydo does with AI falls under the European AI Act as well.
We are taking on a small number of companies as beta partners. We run alongside you for four to six weeks on your own data, and it costs you nothing. Together we measure where things stand up front, so at the end you can see what changed.