phase 2 · sovereign core
Your own core · sovereign AI

Star — your own core.

From a live service grows your own AI brain — on-premise, private, running the company and orchestrating agents from Orbit. AI without adoption is a license without value; Star closes that gap through ownership.

Compass → Launchpad → Orbit → Star · on-premise migration of the same

Learning loop · Star at the corephase 2
STAR SOVEREIGN LLM runs + orchestrates
01What Star is

A core that belongs to you.

The three pillars of Star — why it's more than a rented tool.

Data sovereigntyPillar 1

Your data never leaves your infrastructure. Full control over what the model sees and what it doesn't. Governance, audit, and compliance stay with you.

Your own AI brainPillar 2

An on-premise private LLM that runs the company and orchestrates your agents from Orbit. Not a rented tool — a core that is yours and learns for you alone.

Service → software you ownPillar 3

From a live advisory service grows a core that belongs to you. The end of vendor lock-in. Every month of operation widens a lead that no one can catch up to by buying a license.

02Why your own core

The market is turning. Cloud repatriation is coming.

A large share of CIOs are bringing workloads back from the public cloud under their own control. From August 2026, the EU AI Act becomes enforceable. For regulated sectors — finance, healthcare, public administration — the cloud often isn't an option. Owned data and an owned AI brain become a compounding lead — a moat the competition can't catch up to by buying a license. Every month of operation widens that lead. Whoever starts earlier builds a head start that can't be closed in two years.

08/2026
EU AI Act enforcement beginsRegulation (EU) 2024/1689
every month
The core's lead grows — and can't be bought retroactivelythe moat compounds with operation
0
That much sensitive data goes to an outside vendoreverything stays inside the perimeter
03Architecture

What the core is made of.

Star is neither a black box nor one giant model. It's a four-layer architecture that you run yourself and own entirely. Here's what actually runs inside it — no buzzwords.

Agent orchestrationcontrol

The brain that decides who does what. It takes a task, breaks it into steps, assigns them to the right agent, and watches the result. The same agents you know from Orbit in Microsoft 365 report here and receive instructions from here. Star is the conductor, Orbit is the orchestra.

Custom and fine-tuned models + search over your data (RAG)intelligence

One or more language models fine-tuned to your company, plus a layer that connects them to your documents, contracts, and databases before every answer (RAG). The model doesn't answer from memory — it answers from your current data and can tell you which source it drew on. No hallucinations about what's in your policy.

Compute and storagefoundations

The hardware it all stands on: GPU servers to run the models and storage for your data and its indexes. It sits in your data center or in your private section of a data center. Data flows in, gets processed, and never physically leaves your perimeter. This is the difference between owning and renting.

Policy gateway (governance)control

The layer every request and every response passes through. It watches who has access to what, what the model may and may not do, and writes an audit trail — who asked what and what they got. This is the layer compliance, your auditor, and the EU AI Act will demand. Without it, AI has no place in a company.

Your data stays inside

Star: a sovereign core, data at home.

The loop runs inside the company and feeds into your own Star core. Switch modes and compare: a sovereign core keeps sensitive data at home, the public cloud lets it leak out.

Your company Company data and apps Contracts Finance R&D and production HR and knowledge Projects Knowledge of the company Your space in the data center private AI cloud STAR your own AI core Agents and apps Models and search Compute and storage Policy gateway Public cloud someone else's platform Knowledge at the vendor
Sensitive data stays at homeThe loop learns for the companyOnly generic tasks go out — through the policy gateway
Sensitive data goes outKnowledge accumulates at the vendorAudit and control end at the company boundary
Public cloud AI

Great start, shared rules

Fast deployment and top-tier models — the right choice for generic tasks. But with sensitive data, knowledge of the company stays outside it and control ends at the platform boundary. In regulated sectors, that's a limit.

Your own Star core

Slower start, lasting advantage

On-prem or private cloud. Data stays at home, the loop learns for the company — not for the vendor. Every month of operation widens a lead the competition can't catch up to by buying a license.

04Deployment

Where Star runs.

Star isn't a box in someone else's data center that you rent. You choose where the core sits and who has physical access to the hardware. Three options, one principle: the data stays with you.

On-premise — your own hardwaremaximum

The core runs on servers in your server room, behind your firewall. Data never leaves the building, and only your IT has access to the hardware. A fit for companies where data physically may not go out — finance, healthcare, the public sector, defense.

Private cloud — your space in a data centerbalance

A dedicated, isolated space at a European provider that belongs to you alone — no shared tenant, no shared compute. No investment in your own hardware, while control over the data and the model stays on your side. For companies without their own server room that still don't want the public cloud.

Hybrid — sensitive at home, the rest scalesflexibility

You keep the core and sensitive data on-premise, and offload spikes and less sensitive tasks to a private cloud. You decide which data may cross the perimeter and which may not. For companies that want both sovereignty and the ability to scale at peak times.

05Compliance

Sovereignty in practice.

Sovereignty isn't a slogan, it's a list of properties that hold up before an auditor and a regulator. Here's what it means concretely for regulated sectors.

Data residency — data won't leave the perimeterresidency

Your data, queries, and derived outputs stay inside your perimeter and within the EU. Nothing is sent to an outside API, nothing is logged outside your infrastructure — you have every byte under control.

EU AI Act — ready for August 2026regulation

Enforcement of EU AI Act obligations begins in August 2026 and will touch even where and how your model runs. Your own core gives you the transparency and control a rented cloud won't offer — you can demonstrate compliance, not just promise it.

Audit trail and governancedemonstrability

Every query, every answer, and every access to the model is recorded — who, when, what they asked, and what they got. In an audit or an incident you have a complete trail, not a black box.

RBAC — who sees whataccess

You control access to the core's data and capabilities by role: a salesperson doesn't see payroll, a contractor doesn't see contracts. The model respects your permissions, it doesn't bypass them — sensitive data doesn't reach anyone who isn't entitled to it.

No vendor training on your dataownership

Your data never trains an outside model and never improves an outside product. What the core learns from your operation stays your competitive advantage — a moat that grows for you, not for the vendor.

The path to your own core

Compass → Launchpad → Orbit → Star.

Star is the destination of the journey, not the start. Each layer prepares the ground for the next — and Star is the final distillation: your own AI brain that the company truly owns.

CompassNavigation LaunchpadLaunch pad OrbitLiftoff · agents StarTo the starsyour own core
today
01 · step 1
Compass
discovery · map
  • Maps how the company really works
  • Pinpoints where AI earns the most
  • Surfaces recurring processes
today
02 · step 2
Launchpad
training · adoption
  • Teaches the team to use AI tools
  • By role, on real tasks
  • Captures what recurs
today
03 · step 3
Orbit
agents · 24/7
  • A team of agents tailored to your processes
  • Natively in Microsoft 365
  • A human at the decision
phase 2 · journey's end
04 · step 4
Star
sovereign LLM · core
  • The company's own operating core
  • Data stays and decides
  • For maturity, regulation, and scale
You're here. The company's own core.
06Vision · endgame

An AI brain the company truly owns.

Companies that go through Compass, Launchpad, and Orbit have the foundation for AI to stop being a cost item and become a strategic asset. Star is the endgame: an AI brain the company owns, that learns for it alone and widens its lead every day. Not a rented model with a data dependency on the vendor — a core that grows with the company and stays, even when the market changes. That's a difference that can't be closed in two years. And that difference opens a window today — for those who start.

Star isn't a tool you rent. It's the company's brain that you own — it learns for you alone and with every month of operation widens a lead no one can catch up to by buying a license.

Enter · phase 2 of the loop
07Objections

Frequently asked questions.

What companies considering their own core ask — and honest answers, no sugarcoating.

01Is Star real, or just a vision?+
Let's be honest: Star is phase 2. It's the destination the whole loop leads you to — not a box we'll deliver to you today. What's real right now: Compass maps your company, Launchpad trains your people, and Orbit deploys agents 24/7 in your Microsoft 365. We build your own sovereign core once you have enough processes, data, and reasons to have it. We don't promise Star for next week — we build the path to it that's worth walking.
02How much does it cost and how long does it take?+
We can't seriously give you an exact figure here, and we won't make one up. Both the price and the timeline depend on what Compass uncovers — how many processes, how much data, what regulation, what hardware you already have. It proceeds in phases: first discovery and adoption (a matter of weeks), then agents in Microsoft 365, and your own core only as the last step, for companies where it makes economic sense. Each phase has its own budget and its own return — you don't pay for the endgame up front.
03What if we don't have our own data center?+
Most companies don't, and it's not an obstacle. Star can run in your private section of a data center or in dedicated hosting within the EU, where the hardware and data stay separate and under your control. The difference from the public cloud isn't where the servers sit — it's that the data is yours, the model is yours, and no outsider can see into it. We pick the specific option based on your regulation and budget, not on what we happen to have in stock.
04What about security and data?+
That's the whole reason Star exists. Your data is processed inside your perimeter and physically never leaves it — it doesn't leak into the public cloud, where you don't know who is learning what from it. Every request passes through a policy gateway that controls access and writes an audit trail. For finance, healthcare, and the public sector, this is a requirement, not a luxury — and with the EU AI Act, enforceable from August 2026, a requirement your auditor will demand too.
05What happens to our agents from Orbit?+
You won't throw them away — quite the opposite. The agents running for you today in Microsoft 365 connect to your own core once Star is deployed. Instead of each one running on its own on an outside cloud, a single sovereign brain over your data starts to run them. The work from the Orbit phase isn't rewritten — it becomes the foundation that Star orchestrates. The path is designed so that nothing gets thrown out.
06Why start now?+
Because a lead can't be bought retroactively. The moat — your data and process advantage — grows every month you actually use AI in the company; the sooner you start, the bigger the gap between you and the competition. On top of that, the EU AI Act from August 2026 turns governance into an obligation, and a strengthening trend of pulling data from the public cloud back home. Starting doesn't mean building Star today — it means taking the first step through the loop that ends with you owning it.
Phase 2 · contact

Let's talk about the future.

Star is the vision the loop leads to. We'll show you where your company stands today and what a realistic path to your own core looks like — no shortcuts and no unrealistic promises.