SCARLET WOLF / SERVICES

Your company AI, deployed and operated on your own server.

Your team wants AI, your data must stay home. Scarlet Wolf audits your current usage, deploys a complete AI on your own server and operates it day to day: you get the gains, nobody else gets your data.

DELIVERY REMOTE · FR / EU FIRST CALL 30 MIN, FREE YOUR CONTACT THE FOUNDER REPLY WITHIN 48 BUSINESS HOURS
01

The problem is not adopting AI, it is staying in control

In most SMBs, AI is already there: personal ChatGPT accounts, customer data pasted into free tools, no rules. Banning it does not work. Taking back control does.

INVISIBLE USAGE

Your team already uses it

On personal accounts, outside any oversight. Every question asked can carry a quote, a contract, a customer file.

RESOURCES

Nobody to run it

Deploying and operating an internal AI takes skills most SMBs do not have in-house. That is exactly the service: you need nobody on staff for it.

02

Four offers, one backbone

Depending on where you start: understand, deploy, or delegate operations. In all three cases the architecture is the same: the AI runs on a server that belongs to you.

AUDIT

AI sovereignty audit

A review of your actual AI usage, a map of what leaves the company and through where, and an action plan. Deliverable: a written report and a priced roadmap. The entry point.

DEPLOYMENT

Sovereign deployment

A complete AI deployed on your server, yours or a European one opened in your name: assistant, knowledge base connected to your documents, team accounts, training. Built on Gungnir, our platform with public code.

OPERATIONS

Managed AI

We operate your instance over time: updates, backups, monitoring, support and evolutions. You use it, we handle the rest.

EXPERTISE

By the day, for technical teams

You already have a team and need expert hands: sovereign AI deployment, CPU inference optimization, retrieval and RAG. Expert days on your infrastructure, and the method is public, judge for yourself: our embedder fine-tuning campaign, failures included.

03

How it works

STEP 01

A 30-minute call. Free, by video. Your context, your constraints, and some answers already. If we cannot help you, we say so on that call.

STEP 02

The audit. Your actual AI usage, what leaves the company, what should run locally. A written report and a priced proposal: you know where you are going before committing to anything.

STEP 03

The deployment. Installation on your server, your documents loaded into the knowledge base, team accounts, training. You own the instance from day one.

STEP 04

Operations. Updates, backups, monitoring and support over time. Or on your own if you have the technical skills: all the code is public.

04

We live what we sell

Asking a provider where its own tools run is a good test. Here is our answer.

OUR INFRA

Zero US cloud

Code, continuous integration, image registry, website: the entire Scarlet Wolf infrastructure runs on our own servers, in Europe. We left GitHub for that.

FLAGSHIP

Gungnir, in production

The platform we deploy for you is the one we use ourselves every day. Public, verifiable code.

DEPTH

A journal, not brochures

CLOUD Act, GDPR, open-source LLMs: we publish sourced analyses on the subject. Judge the depth before talking to us.

If Scarlet Wolf disappears tomorrow, your AI keeps running. That is what ownership means. → FULL REVERSIBILITY
05

The questions we get asked

How much does it cost?
Three components: your server (a few dozen euros per month for most SMBs), your model usage at cost price (your keys, or free open models running locally), and the service itself, quoted after the audit. No per-seat license, no hidden subscription.
We have no server and no technical team. Is that a blocker?
No, it is the most common case. We size and open a European server in your name: you own it, we work on it as a provider. No technical skills are required on your side.
What if we stop the service?
You keep everything: the server is yours, so is the data, and the platform code is public. The instance keeps running without us. Reversibility is not a contract clause, it is the architecture.
Which AI models do you use?
The ones that match your confidentiality requirements: open models running fully locally (Mistral, Llama and others), or the major API models with your own keys for what is allowed to leave. The architecture is multi-model, you switch whenever you want.
Where does GDPR fit in?
Self-hosting removes most of the problem: no transfer outside the EU, no hidden processor in the chain. We hand you the technical elements for your record of processing activities, and data export and erasure are built into the platform.
SERVICES · FIRST CALL FREE

Thirty minutes to know where you stand.

Your context, your constraints, concrete answers. And if we cannot help you, we say so on that first call.