Stone and glass house behind a dry-stone wall in bare highland hills under an overcast sky

SECURITY

European by structure, not by hosting region.

BlackMountain is a European company with European ownership, operating entirely within the EU. No US parent, no non-EU investors, no foreign jurisdiction reaching into your data.

European ownership – no investors or parents outside the EU
100 %
Application, storage and processing run in the EU (Scaleway, Paris)
EU
Of your content used to train models
0
Deployment options, from EU multi-tenant to on-premise
4

“Hosted in the EU” is not the same as European. Many AI vendors run EU data centres while remaining under non-EU corporate control. Sovereignty is a question of ownership and jurisdiction – not of hosting labels.

  • FIG 0.1

    Jurisdiction follows ownership

    A company under non-EU control can be legally compelled to disclose data, wherever it is stored.

  • FIG 0.2

    Model providers add exposure

    AI processing often leaves the EU even when the application does not.

  • FIG 0.3

    Compliance has to be provable

    Auditors and procurement teams need documentation, not marketing claims.

WHAT MAKES US SOVEREIGN

Four things you can verify.

100% European ownership

BlackMountain has no investors or parent entities outside the European Union. No foreign jurisdiction applies to our corporate structure.

EU-only operations

Application, data storage and processing run within the European Union. Our Datacenters are operated by the European provider Scaleway and are located in Paris.

Controlled model access

Models are operated inside a governed EU environment. You decide which providers are enabled for your organisation.

Your knowledge stays yours

We use an Enterprise API – your content is never used to train models. The know-how of your workforce stays within your company instead of leaking into external systems.

HUMAN-LED AI

AI that works with people, not around them.

Human judgment stays in control

AI can research, summarise, draft and recommend — but critical decisions, approvals and permissions remain with people, with governance, roles and access control built in.

Ask Human instead of guessing

When information is missing or a request is unclear, AI should not invent an answer. With Ask Human, the system can ask for clarification before continuing.

Designed for collaboration

People and AI work in the same workspace, with shared context, connected knowledge and visible workflows. This makes AI part of how teams work – not a separate tool on the side.

A platform built for quality

Purpose-built features ensure you keep long-term control over your business processes — not the AI. Because the best results happen when you stay in the lead.

GOVERNANCE & CONTROL

Security you can administer, not just trust.

Permissions, data protection and oversight are built into the platform – administered by your teams, traceable for your auditors.

  • INPUT Request → ROLE Organisation, ROLE Workspace, ROLE Team → OUTPUT Permission-aware Answer

    ACCESS

    Every answer respects who is asking.

    Roles on organisation, workspace, team and knowledge level. AI only sees what the requesting user is allowed to see – nobody gets more through AI than in the original system.

    Explore Control Tower
  • INPUT Prompt & Files → CHECK Detect PII, CHECK Mask, CHECK Route → OUTPUT Protected Request

    DATA PROTECTION

    Personal data is caught before it reaches a model.

    PII protection detects and controls personal data in prompts and documents. For the most sensitive information, Local AI keeps processing in your own building.

    Explore Local AI
  • EVENT Agent Run → STEP Ask Human, STEP Approve, STEP Log → OUTPUT Audit Trail

    OVERSIGHT

    People decide. Everything is traceable.

    Ask Human keeps people in the decision path, and activity, access and model usage are recorded for internal and external review.

    Explore Agent Studio

CERTIFICATIONS & DOCUMENTATION

Documented, not asserted.

ISO 27001

IN PROGRESS

GDPR compliance and data subject rights

DOCUMENTED

DPA available for all customers

AVAILABLE ON REQUEST

Subprocessor list published

PUBLISHED

Independent penetration testing

DOCUMENTED

Encryption at rest and in transit

DOCUMENTED

DEPLOYMENT OPTIONS

Choose how 48° runs for you.

01 / STANDARD

EU Multi-Tenant

Standard setup, fully EU-operated.

02 / ISOLATED

Dedicated Environment

Isolated instance for higher assurance requirements. From 1.000 Seats.

03 / ENTERPRISE

Your own cloud

Deployment into your EU cloud account. Enterprise.

04 / REGULATED

On-premise

For regulated environments with strict residency rules. Enterprise.

LOCAL AI / NEW

For data that must never leave your building.

Run open-weight models on a Mac mini or Mac Studio in your own office or data centre, connected to 48°. Workspaces you mark as local are answered only on your hardware – everything else keeps using the EU platform.

Mac mini on a desk with display, keyboard and mouse, seen from above

BUSINESS MODEL / TOKEN INCENTIVES

Aligned incentives for responsible AI usage.

BlackMountain does not build its business on token markups. That means our incentive is not to make customers consume more tokens — but to help organisations use AI efficiently, securely and with measurable value.

No token margin incentive

We do not earn more by making your AI workflows unnecessarily expensive.

Built to reduce AI waste

Our platform can optimise how context, models and workflows are used — reducing unnecessary token consumption where possible.

Right model for the right task

Not every task needs the largest or most expensive model. Multi-model access helps teams choose the right model for the job.

Reusable workflows

Instead of repeating long prompts and manual context setup, teams can create reusable assistants and workflows that standardise efficient AI usage.

Cost visibility through Control Tower

Usage and cost are visible across teams, models and workflows — so AI adoption can be managed instead of guessed.

EVALUATION FRAMEWORK

Five questions worth asking — including us.

  1. 01

    Who owns the company, and which jurisdiction applies?

  2. 02

    Are there investors or parent entities outside the EU?

  3. 03

    Where does model inference actually happen?

  4. 04

    Which subprocessors are involved, and are they disclosed?

  5. 05

    Can permissions from your source systems be enforced inside the AI?

We answer all five in writing. If a vendor cannot, hosting location is not the reassurance it appears to be.

FAQ

Straight answers to the questions that matter.

READY WHEN YOU ARE

Bring AI into your organisation without giving up control.

Talk to our security team or review our documentation before you decide.