The strongest AI in the EU.
European AI. European values for European data.
Susi runs on Mistral by default — the strongest model the EU has to offer.
More on this ↓The tool for Teamwork with AI.
The EU's strongest AI. Little effort, full benefit.
One project, separate chats — and everything still in one place.
More on this ↓When the AI carries your name.
Full branding, even for small teams.
On your own subdomain, with your own library.
More on this ↓Next-level meeting minutes.
For example: 25 minutes of video, transcribed in under a minute.
Recordings up to two hours. Automatic speaker detection, and you can question it afterwards.
More on this ↓Based in Upper AustriaServers in the EULanguage model: Mistral, ParisCancel monthly
Surprisingly Useful Speech Intelligence
The EU's strongest AI.
European AI. European values for European data.
Susi runs on Mistral by default — the strongest model the EU has to offer. Other models can be used on request. When the model gets better, you are working with it automatically, shortly after release. Writing, summarising, taking minutes, searching your own knowledge — what Europe builds is enough for that. You pay for use, not for upkeep.
Request a demo- The EU's strongest model
- Developed in Paris
- Run in European data centres
- Your content is not used for training
- No server for you to run
Teamwork with AI.
One project, separate chats — and everything still in one place. Hand out tasks, ask for status, nothing gets lost along the way.
- Say it once. Everyone knows.
- Everyone picks up their own task.
- Status on demand.
Susi passes on what you tell her and keeps track of what gets reported back. Questions only work inside the shared project, and only within your own team. How the project is run is decided by the project lead.
Request a demoMore on teamwork →One project, separate chats — and everything still in one place.
When the AI carries your name.
Full branding, even for small teams. Your name, your look, your address — on your own subdomain, for instance.
Your own library with your company knowledge, filled by you and kept current by you. Clear processes, high efficiency. Fair, affordable setup costs instead of a development project.
Request a demoMore on your own AI →Next-level meeting minutes.
For example: 25 minutes of video, transcribed in under a minute — from upload to the finished transcript.
Recordings up to two hours. Even when the line is slow. We are fast and we store no video. Automatic speaker detection, and afterwards you can ask: "What was decided about the budget?"
Request a demoErnaThe brochure is drafted, I still need the prices.
HansThose only come after the costing.
TeufelThen we'll postpone the budget to next week.
What you were probably looking for
Which AI is there from Europe?
Several. The strongest one in the EU comes from Mistral in Paris — SUSI runs on it by default. The conversation itself is the one you already know: you type, SUSI answers, you refine. Anyone who has written with an AI before needs no training.
The difference is everything around it. Where the data sits. Who manages the accounts — you do, without us. Whether your own company knowledge is part of it. And whether you can walk away at the end of the month without a discussion.
I want our own AI — can SUSI do that?
Yes, if you mean what most people mean by it. No, if you mean your own language model on your own hardware. The difference matters, because it decides what it costs and how well it works.
What "your own" means in practice. Anyone asking for their own AI almost always wants three things: it should know our documents. It should carry our name and be reachable at our address. And our data should not end up somewhere. All of that we set up for you — your library with your company knowledge, your colours and your logo on your own subdomain, servers in the EU, and your content is not used to train models.
What "your own" can also mean. Some mean a model running on their own hardware in-house, trained on their own data. That is technically possible. Only:
- The models you can sensibly run on affordable hardware are many times smaller than the large ones. The answers are accordingly.
- Buying it is the smaller part. Operation, power, updates and someone who knows the subject keep running.
- A trained model is a still image. If your product documents change, you have to retrain.
- At the next leap in models you start over.
One advantage remains. With your own hardware the data stays in the building, under your own control. That is a real argument — it just applies to very few today. Anyone using Microsoft 365, SharePoint, OneDrive, Dropbox or any other cloud service has long since stopped keeping their data under their own roof. Exceptions do turn up, though not particularly often.
Why we build it differently. Training teaches a model a tone, not knowledge — for knowledge it is the least precise route. The library looks things up in your documents with every question and names the document and the passage. You can check what the answer rests on. Updating means: swap the file, done. That is more precise, it is cheaper, and you keep control over what is in there. Nothing to set up, nothing to maintain, nothing to upgrade — and when the model gets better, you are working with it automatically, shortly after release.
The full story is on Your own AI for business.
Can an AI work with our own documents?
Yes. And without any training.
This is a common misunderstanding: that to tune an AI to your own company you have to train it on your own data. That is expensive, slow and delicate in data protection terms — SUSI does not do it. Instead your material is searched at the moment of the question and the matching passage is used as the basis for the answer.
That has three consequences which matter more day to day than the technology behind them: a new document takes effect immediately. A removed document stops taking effect immediately. And every answer can be traced back to its source passage.
What does an AI licence cost per employee?
€14.90 net per month for the single licence. Cancel monthly, no minimum term, no setup fee. Every account has a usage allowance which the employee only ever sees as a percentage, never as an amount — nobody should have to weigh up whether a question is worth asking.
What Company, Org and your own environment cost is further down.
Is it available as white label on our own domain?
Yes, at any size — from a single workstation to a company with many teams. Your name, your logo, your colours, your address. SUSI's name appears nowhere.
Plus your own library that knows nothing but your company knowledge. Setup costs are fair and affordable, not a development project. What you contribute: an address, a name, a logo and a colour.
What SUSI is used for day to day
Six situations from the working day — and what SUSI does in them.
Always the same questions
"How do I apply for this? Who is in charge?" — and it is always the same three people who know.
In your own environment you build a knowledge base for the company — policies, manuals, forms, your own website.
It is searched automatically with every question, without anyone having to select anything. The answer names the document and the passage.
What did we agree?
Two weeks away, and nobody remembers what was agreed.
When a project is shared with a team, every conversation in it becomes part of a shared memory. The search space is the project, not the individual person.
"What did we agree on for the interface?" "What has happened since the start of the month?" Anyone coming back from holiday asks instead of reading.
Nobody writes the minutes
Everyone was there, nobody took notes.
Upload the recording, audio or video, up to two hours. Out comes a transcript with timestamps and separated speakers — as Word, PDF, text or Markdown.
The point comes afterwards: one click moves the finished minutes into your knowledge base. From then on the meeting can be questioned — "What was decided about the budget in June?" — with a reference to the passage. And the whole thing is faster than you have probably ever seen.
The chat is full
The conversation gets long, the answers get vaguer, and at some point nothing works at all.
That is not a plan limit. Every language model has a context window — once it is full, the conversation ends, or the beginning quietly drops out.
One click and SUSI files the conversation as a document in your knowledge base. A new conversation starts empty — and still knows what was in the old one.
"I don't forget anything. I just tidy up — and remember where I put it."
Texts that sound like you
The quote, the newsletter, the customer reply — it just doesn't sound like you.
Quotes, product copy, replies to customer enquiries. SUSI suggests, you decide — and refine as you go: "too formal", "shorter", "no hedging".
Word, PDF and Excel are created right in the conversation and are ready to download. What SUSI can do in detail
The new colleague starts Monday
First week, two new colleagues, and the person meant to train them has no time.
SUSI puts together a summary from the company's own material — processes, responsibilities, forms — and hands it to the two of them.
Ask more and you get more: the knowledge base sits behind it, with a reference to the passage. And the onboarding does not repeat itself — what is in the material once holds for the next newcomers just the same.
Four ways to work with it
SUSI is used in very different ways. Pick whichever comes closest to your situation.
The one-person business at home
When you work alone, there is no colleague to take a quick look.
- Write quotes, emails and copy — in your own tone
- Have contracts, official letters and long PDFs summarised
- Record meetings and calls and have them transcribed
- Research, translations, images
- Your own documents in the library — the answer knows the passage
- On request with your name and your own subdomain
The small-business office
Two people are enough. A team, a law firm, a practice — everyone needs the same answers, from an internal AI that knows the company's knowledge.
- Shared project: hand out tasks, ask for status
- Company knowledge in the library — policies, templates, manuals
- Ready-made assistants for recurring tasks, set up once
- Minutes the whole team can read back
- New colleagues get themselves up to speed
- On request with your name and your own subdomain
The training environment
Anyone teaching AI needs something you can show — not something you have to explain. And you need practice accounts — which have to carry over from one course to the next.
- Simple user management, ideal for several courses a month
- Example prompts, one click for all participants
- Compare different models in the same tool
- Demonstrate minutes and teamwork live
- An environment with your name instead of third-party accounts
- Everything stays in the EU — the practice data too
The large company with many teams
Many departments, many projects — one line.
- Separate projects per team, shared libraries per division
- Internal AI on your subdomain, in your colours
- You manage accounts and teams yourselves
- Your own database on request
- Data processing agreement, notification within 48 hours, right of audit
White label comes in every size. From a single workstation to a company with many teams — your name, your look, your subdomain. It does not depend on how many of you there are.
European AI — and what that means exactly
Servers in the EU, a European CDN, the language model from Mistral in Paris. No Google services, no US cloud hosting. Where there are exceptions, they are stated here — not in the small print.
Where is our data?
Based in Upper Austria. Servers in the EU, for the website and for SUSI. The language model comes from Mistral in Paris. A European CDN. No Google services — no fonts either, and no analytics tool. Only technically necessary cookies, no tracking, no advertising. The protection against automated sign-ups runs without third-party services.
Loading this page sends not a single request to a third-party server. If you want to check: open the developer tools, "Network" tab, reload.
Your conversation histories are kept with us, not with the model provider. We deliberately forgo the obvious cost saving that storing them there permanently would bring. Your content is not used to train models — that is in the contract terms.
Tenant-separated, data protection under EU law. Your own database is available on request.
Where is the boundary?
Image generation runs in the USA, with a provider that does not store your requests. Web search runs through Staan¹, a European search index.
"When I draw, I look over the fence. For everything else I stay at home."
¹ Staan has a fallback to the USA when there are no results on a topic.
What about the US CLOUD Act?
The US CLOUD Act obliges American providers to hand over data in their possession or under their control — regardless of where it is stored. What matters is therefore not where the server stands, but who holds legal and actual control.
For a European provider, Art. 48 GDPR applies: an order from a third-country authority can only be recognised in the EU on the basis of an international agreement. In 2019 the European Data Protection Board and the European Data Protection Supervisor concluded that, absent such an agreement, an EU provider generally cannot lawfully comply with a direct US disclosure request.
That is not a promise that nothing can happen. It is the difference between a provider caught between two legal systems and one that is not. The US CLOUD Act in detail
Why does it never say "GDPR compliant" anywhere?
Because we cannot give a solid assurance of it for our own operation — and because it does not rest with us alone. Whether a processing operation complies with the GDPR depends on what you use SUSI for and what data you put in. That is decided at your end, not ours.
There is also a boundary we do not hide: where EU endpoints exist, we use them. They do not exist everywhere: image generation runs in the USA, with a provider we chose because it does not store your requests. For web search there is a European route: when SUSI looks something up on the web, she asks Staan¹, the European search index from Qwant and Ecosia, with data centres in France.
What goes out is less than it sounds: the search runs through our server, with our IP address. What arrives at the search service is a search phrase the model has formed — no IP of yours, no browser or device details, not your conversation, not your documents and nobody who asked.
What we do control, we have decided: delivery through BunnyCDN in Europe, not through Cloudflare. The protection against automated sign-ups runs on our own server, not through a captcha or Turnstile. No fonts from third-party servers. And no tracking, in any form.
Instead we state what can be checked: where the data sits, who processes it, on what basis, and what the contract says. Data processing agreement, breach notification within 48 hours, a right of audit on your side — including by an external auditor — deletion including the search index with no recycle bin, on request with written confirmation. The separation between tenants and between individual users has been verified with targeted attack testing, not merely asserted.
And for completeness: a Google search — which most people run several times a day without thinking about it for a second — does not stay in Europe either. This is not about purity. It is about you knowing where you stand.
¹ Staan has a fallback to the USA when there are no results on a topic.
What is in the contract?
Data processing agreement, incident notification within 48 hours, right of audit, deletion with written confirmation. Full chat export at any time — what you wrote belongs to you.
SUSI rocks.
Surprisingly Useful Speech Intelligence
An AI that carries your company's name.
Doesn't have to be expensive.
It starts with a licence: buy it, the credentials arrive, off you go. No minimum term — and at the end of the month it stops just as easily as it started. No setup fee for the standard model, a fair one for white label.
White label — your configured AI on your domain
Your own name, your own look, your own address — and a library that knows nothing but your company knowledge. SUSI's name appears nowhere.
Three invented examples. What she is called, what colour she wears and the address she answers at is up to you.
How getting started works
Three ways in — depending on whether you start alone, with a team, or for the whole company at once.
- 1
Single licence — right away
Buy it in the shop, access follows once payment arrives. Cancel monthly, no minimum term, no setup fee.
- 2
Your own environment — one conversation and it stands
A short conversation about what you need, then a quote. You contribute an address, a name, a logo, an icon and a colour — and a voice recording for your own voice if you want one. We set up everything else; your environment stands the same day. Filling it with your knowledge you do yourselves: importing the website and uploading documents is operation, not project work.
- 3
One division first, then the company
We recommend starting with one division and one clearly defined project rather than the whole company. After four weeks you know where the benefit actually lies for you — and usually it is not where you would have guessed.
We sell to businesses only; to buy in the shop you need a VAT ID. Small businesses without a VAT ID and associations, please email us — that works too, just not automatically.
From the same workshop: our WordPress plugin AI Image check&mark marks AI-generated images in a machine-readable way.