eutonomous is still being built — nothing is live yet.Interested? Get in touch
In development — nothing is live yetEU infrastructure, EU law — no CLOUD ActOne account for everything

Sovereign AI, from the browser to the GPU.

AI workspaces that do the work when you ask, GPU boxes for the heavy runs, apps that run on your users' own devices, and an AI API that is secured on every request — under one eutonomous account.

Servers in Europe aren't enough. Everything that touches your data runs on EU infrastructure, operated by European companies, out of reach of the US CLOUD Act.

See the productsGet in touch

The portfolio

Four ways to build with AI

Each one stands on its own, with its own site. Use one, or move between them as the work grows — the same eutonomous account signs you in to all of them.

in development

eutonomous AI workspaces & sandboxes

Your own sandbox — ask for the work, or take the controls.

A personal compute sandbox that you control: files, data and running apps stay in it between sessions, and it sleeps when idle so you pay only for what you use. There are two ways into the same sandbox — anything the agent builds, you can open by hand, and anything you set up, the agent can use.

The AI workspaces chat: suggestions such as making sense of a spreadsheet or turning notes into a PDF

Ask

AI workspaces

Describe an analysis, a report, a chart or a working page in a chat, and the agent does the work in your sandbox. For people who do not code, and for those who would rather not.

See the preview

Creating a sandbox on the platform: stacks such as Next.js with Postgres, FastAPI with React, Django or WordPress

Do it yourself

Sandboxes on the platform

Assemble the sandbox yourself — frameworks, databases, admin tools and editors — then work in VS Code or a terminal in the browser, or keep vibe coding with Mistral inside it.

See the preview

in development

Rixi compute boxes

Your project on a GPU, in one command.

Rixi ships your code and its exact pixi environment to a GPU or CPU box, runs it and streams the output back — on a box we rent for you in an EU-owned cloud, or on your own hardware on premise.

  • Rented boxes on EU-owned clouds only
  • Or bring your own on-premise box
  • Your traffic goes straight to the box — not through us

See the preview

The Rixi page: one pip install and one rixi run command send a project to a GPU box
in development

Client-side runtime

Apps that run on your users' devices.

Write Python apps that execute in the browser and reach your own databases through an encrypted tunnel. No app servers to run, and no data copied to someone else's cloud.

  • Streamlit, Marimo, JupyterLite, PyScript and PyLevate apps
  • Data stays in your infrastructure
  • Secrets never reach the browser

See the preview

Creating a project in the client-side runtime: Stlite, Marimo, JupyterLite, PyScript or PyLevate
in development

Inference API

EU models behind one endpoint, secured on every request.

OpenAI- and Anthropic-compatible, so the coding agent you already use works unchanged. Tokens cost what the provider charges, and personal data, risky prompts and hidden injections are dealt with on the way.

  • Every model at a European provider
  • Personal data replaced before the model sees it
  • A signed record you can verify offline

About the API

The governance dashboard: requests, blocked and flagged prompts, personal data replaced, by day and by rule

Screenshots are of the products as they are being built. The preview sites are where we test them; none of them is open for customers yet.

Still being built

Nothing is live yet — but we would like to hear from you

eutonomous is in development. If one of these products fits what you are working on, tell us: early conversations decide what ships first, and you will be the first to know when it does.

Get in touch

How it fits together

Start small, scale up, ship

The products line up along the way a project grows — and none of them requires the others.

01

Work it out in a workspace

Ask for a first version and refine it in conversation, or open the workspace and carry on by hand. It sleeps between sessions.

02

Burst to a GPU when it is heavy

Training run, batch job, big model: send the same project and environment to a Rixi box for as many minutes as it needs.

03

Ship it the way it fits

As an app that runs in your users' browsers on your own data, or behind the inference API with security on every call.

What they share

Sovereign by construction, everywhere

EU infrastructure, EU law

Location is only half of it: a US-owned cloud in Frankfurt still answers to the CLOUD Act. Models, workspaces and boxes run with European companies on their own EU infrastructure — the sovereignty ledger documents each one.

One account

A single eutonomous sign-in for every product, on our own identity service in the EU — not a US login provider.

Not a black box

Open your workspace, read every file, take your code with you. The Rixi client and SMCP are open source; malgra, the gateway underneath, will follow.

Honest pricing

Tokens through the inference API cost exactly what the provider charges, with no margin. Workspaces and compute are prepaid and billed for what you actually use.

Under the hood

One gateway guards the model calls

Inference requests pass through malgra, the gateway we wrote and intend to release as open source. You never need to know it is there — but it is written down.

What runs underneath