Skip to content
Wegenty
Our principles

Sovereign by default. Open by principle. Human by design.

Three commitments shape everything Wegenty builds. They are not posture — each is tied to a mechanism you can point at. The product is not retrieval; it is governance, and the local consultant is the user interface. Here is what each principle means, and how each one is proven.

Knowledge stays in the building Open core, one closed shield A person is always in the loop
The spine

A principle is only worth the mechanism behind it

Most vendors promise trust. We tie each commitment to the specific architecture that earns it — so the claim is falsifiable in a demo, not a mood.

01

Sovereign by default

Knowledge stays under the company’s control. Nothing ships to a foreign cloud by default.

02

Open by principle

The entire core is open source. Only Cloud Shield, the compliance umbrella, is closed.

03

Human by design

A trusted local consultant runs it, and a person is always in the runtime loop.

Principle one

Sovereign by default

Sovereignty is the default state, not a paid upgrade or a checkbox you have to find. Your knowledge belongs to your building, and it stays there unless a person decides otherwise. A cloud CX agent makes you ship your knowledge to a foreign cloud just to switch it on — Wegenty inverts that bargain.

The proof is structural. One locally-hosted LLM serves many knowledge bases, so the brain runs where the data lives. The public customer agent is only ever a governed projection of a private superset, with sensitivity labels and per-base policy deciding what a customer can ever see. Projection is one-directional by construction — there is no path back from the public agent into the internal corpus.

How it’s proven
Local LLM, many knowledge bases

One brain runs on your own appliance, sealed and pure-local by default — or pinned to an EU region if you choose. Public exposure is additive, never the price of switching it on; the model never has to phone a foreign cloud to answer.

Knowledge stays in the building

Bases start closed and default-private. Processing is local, lawful and demonstrable under DSGVO; nothing leaves by default.

One-way governed projection

The internal base is the superset; the public agent is a projection of it, never the reverse. A customer query can only reach what policy explicitly promoted.

Petra · SME owner

“Your knowledge, working for you — without leaving the building.”

How it’s proven
Open-source core

The platform, the governance and the connectivity are open. You can read it, run it, and self-host the whole stack on your own servers.

Only Cloud Shield is closed

A single closed component: the compliance umbrella carrying live regulatory posture, the audit trail and certification — the part no solo operator wants to build or self-insure.

Fork everything but the liability

The open core can be copied; the liability cannot be forked. Certification flows only through Cloud Shield, and the relationship only through your consultant.

Jan · developer

“Open core, sovereign by default — fork everything but the liability.”

Principle two

Open by principle

Openness is the starting position, not a concession. The entire core is open source — there is no opaque model deciding your customers’ answers, and no lock-in disguised as convenience. You are free to inspect it, extend it, and walk away with it.

Exactly one thing is closed. Cloud Shield is the compliance umbrella that carries the live regulatory posture, the audit trail and certification under DSGVO and the EU AI Act. That is the deliberate boundary: openness where it builds trust, a closed shield where someone must carry liability that a solo operator never could.

Principle three

Human by design

The human is not a fallback bolted on at the edges — they are the design. The canonical SME operator is capable but not a power user, so the easiest interface is not a cleaner dashboard but a trusted local person who handles the hard parts. The consultant is the user interface.

A person is in the runtime too. The collaborative correction loop means that when the agent is unsure it escalates to a human instead of guessing — and that human answer is captured back into the knowledge base, so the system improves precisely where it was weak. Good answers stop dying inside support tickets. And the relationship is human too: the agent acts as a concierge, recognising returning customers through passwordless accounts and carrying their history in a sovereign on-core CRM — while the consultant stays the concierge for the owner.

How it’s proven
Consultant-delivered

A trusted local advisor installs it, curates the knowledge, and sets the projection policy — so a non-technical owner adopts it without an engineer.

Escalation, not guessing

When the agent meets uncertainty it hands off to a person who knows the customer, rather than inventing a confident-sounding answer.

The answer becomes knowledge

The human resolution feeds back into the base. The system learns from the team that corrects it, exactly where it was weakest.

A concierge, not a deflection bot

Passwordless accounts and a sovereign on-core CRM let it recognise returning customers and tend the relationship over time — context that stays on your own core, never a foreign cloud.

Thomas · end customer

“A straight answer, or a fast hand-off to a human who knows you.”

The payoff

Agents you can trust.

Controllable. Explainable. Accountable.

The three principles are not separate promises — they are the same commitment seen from three sides. Sovereign by default makes an agent controllable. Open by principle makes it explainable. Human by design makes it accountable. Together they are why you can trust what it does — see how that holds up against a cloud CX agent, or read the full questions and answers.

Controllable

You hold the reins — sovereign, default-private, and a one-way governed projection.

Explainable

No black boxes — open core, citations, a full audit trail.

Accountable

A person stands behind every answer, and behind the liability.

In short

Questions about the principles

Are these principles marketing, or are they enforced?

Each is tied to a mechanism you can point at. Sovereign is proven by a local LLM and a one-directional governed projection — knowledge cannot leave the building by default. Open is proven by an open-source core where only Cloud Shield is closed. Human is proven by a consultant who delivers the system and a correction loop that escalates uncertainty to a person. A principle without a mechanism is a slogan; these have mechanisms.

If the core is open, what stops a competitor from copying it?

Nothing — and that is the point. You can fork everything but the liability. The open core can be copied, but certification under DSGVO and the EU AI Act flows only through Cloud Shield, and the customer relationship flows only through the local consultant. Distribution and compliance reinforce each other, which is exactly why the model is durable.

How is "sovereign by default" different from a normal cloud assistant?

A cloud CX agent ships your knowledge to a foreign cloud as a condition of working. Wegenty inverts the default: one locally-hosted LLM serves many knowledge bases, the data stays in the building (or in an EU region for the managed option), and the public agent is only ever a governed projection of a private superset — never the reverse.

Why does "human by design" mean a consultant rather than a better UI?

The canonical SME operator is capable but not a power user. For them the easiest interface is not a cleaner dashboard but a trusted local person who handles the hard parts. The human is also in the runtime: when the agent is unsure it escalates to a person instead of guessing, and that human answer is captured back into the knowledge base.

See the principles hold — on your own knowledge.

Try the interactive demo, or book a walkthrough of the governance architecture. Sovereign by default, open by principle, human by design.