Sovereignty

Your data. Your model.
Your hosting.

Swiss-hosted or entirely on your own infrastructure, up to air-gapped. The model follows the sensitivity of the data. No US Cloud Act exposure.

Sovereignty

Data sovereignty for discerning companies.

AIgent is Swiss-engineered and Swiss-hosted. Run the platform in our Swiss cloud or entirely on your own infrastructure, up to air-gapped. revFADP and GDPR compliant out of the box. See on-premise details in our on-premise plan.

Talk to us about on-premise

Swiss data centers or on-premise

Platform and data in Switzerland under the Federal Act on Data Protection (revFADP) and GDPR, or entirely on your own infrastructure via Kubernetes, up to air-gapped. No US Cloud Act exposure, off US hyperscalers.

Model choice per agent

The model follows the sensitivity of the data. The most sensitive data never leaves the building: on-premise model, Swiss-hosted, or international models, chosen per agent.

A complete audit trail

Every interaction is tracked: model, tokens, cost, and time. Data export and account deletion are built in. Designed for revFADP, EU AI Act, and DORA.

Model choice by data classification

The model follows the sensitivity of the data.

You set per agent where processing happens. From an on-premise model for banking secrecy to a powerful international model for non-critical tasks.

In your own house

On-premise model for the most sensitive data. It never leaves your infrastructure.

Banking secrecy · customer data

Swiss-hosted

Sovereign models for internal, confidential business data.

EU or US

Powerful international models for non-critical, non-personal tasks.

Other models

Open source, ChatGPT, Claude, or Gemini, connectable per agent when a use-case requires it.

The most sensitive data never leaves the building.

Two operating models

Sovereign in the cloud or on your premises.

The same product, two ways to run it. You decide where your data and models live.

Swiss Cloud

Usage-based, no per-seat fees

  • Hosted in Swiss data centers
  • revFADP and GDPR compliant out of the box
  • Pay-per-use, you pay for what runs
  • Land with one agent, expand across teams
  • Off US hyperscalers
  • First agent live in under 15 minutes, no credit card
Join the waitlist

Why AIgent

Sovereign and a real agent platform.

ChatGPT, Claude and Copilot are powerful, but bound to US companies and the US Cloud Act. Langdock is model-independent, yet doesn't host in Switzerland and bills per seat. AIgent unites both: the power of a real agent platform with Swiss sovereignty, billed per use.

Feature comparison. AIgent vs Langdock, Claude, ChatGPT Enterprise and Microsoft Copilot
AIgent Langdock Claude ChatGPT Enterprise Microsoft Copilot
Sovereign, off US hyperscalers 1 Yes No No No No
On-premise on your own infrastructure 2 Yes Yes No No No
A real agent platform, not just chat 3 Yes Yes Yes Yes Yes
Model-independent, choice per agent 4 Yes Yes Yes No Yes
Reliable skills over tested code 5 Yes Yes Yes Yes Yes
Pay-per-use, no per-seat fees 6 Yes No No No Yes
Designed for revFADP, EU AI Act, DORA 7 Yes No No No No
  1. Sovereign, off US hyperscalers. Data residency is not sovereignty: even Copilot's Azure Switzerland region runs on a US hyperscaler. Microsoft, OpenAI and Anthropic are US companies under the US Cloud Act; Langdock hosts in the EU on Azure.
  2. On-premise on your own infrastructure. AIgent runs on your own Kubernetes, up to air-gapped. Langdock offers on-premise only from 5,000 seats.
  3. A real agent platform, not just chat. AIgent is built from the ground up as an agent platform with pre-configured agents. The others add agent features onto a chat or developer product.
  4. Model-independent, choice per agent. Langdock is fully model-independent; Claude offers model choice via Cowork and Code, Copilot via Copilot Studio. ChatGPT Enterprise stays OpenAI-only. On AIgent you choose model and processing location per agent.
  5. Reliable skills over tested code. On AIgent, skills are vetted, tested code with reproducible results from a centrally governed catalog. The others can run code or tools but offer no such vetted skill catalog.
  6. Pay-per-use, no per-seat fees. AIgent bills purely per use. Langdock, Claude and ChatGPT require per-seat licenses for their enterprise product. Copilot also offers usage-based options alongside seats.
  7. Designed for revFADP, EU AI Act, DORA. AIgent is built for Swiss data residency and a complete audit trail, addressing the core of revFADP and DORA. The others meet GDPR and ISO and offer some EU AI Act and DORA tooling, but are not designed for these and remain subject to the US Cloud Act.

Row-by-row explanations are listed below the table. The edge is the combination, not any single checkbox. Data based on publicly available information as of June 2026.

Questions & answers

Common data sovereignty questions.

Is AIgent GDPR and revFADP compliant?
Yes. AIgent is revFADP and GDPR compliant out of the box. The platform is built and hosted in Switzerland, and all data resides in Swiss data centers by default, under the revised Swiss data protection act and the GDPR. Data export and account deletion are built in, so data subject rights are met without special processes. Every AI interaction is recorded in the audit trail with model, tokens, cost, and time, making processing traceable for data protection officers and regulators. KPIs are built over agents and tasks, not individual people. The platform is also designed for the EU AI Act and DORA, which eases sign-off in regulated industries such as banking and insurance. For maximum control, you can run AIgent fully on-premise on your own infrastructure.
What does sovereign AI mean?
Sovereign AI means you decide where your data resides, which model processes it, and who can legally access it. Sovereignty comes from the operator structure and the hosting, not from a label on a US provider region. With AIgent this means: operation in Swiss data centers under Swiss law, or entirely on your own infrastructure, independent of US hyperscalers and outside the US Cloud Act. You choose the model per agent, so highly sensitive data uses an in-house model while non-critical tasks may run on powerful international models. A complete audit trail lets you verify every processing step. Data control, model choice, and traceability belong together: only when all three are in your hands is an AI platform truly sovereign.
Does the US Cloud Act reach my data?
Not with AIgent. The US Cloud Act applies based on where the provider is incorporated, not where the data center stands: any US corporation can be compelled to hand over data even when it sits on Swiss or EU servers. Even an Azure region in Switzerland is subject to it, because Microsoft is a US company, and a contractual assurance does not change that. AIgent runs independently of US hyperscalers, in our Swiss cloud under Swiss law or entirely on-premise at your site, up to air-gapped. That removes the Cloud Act’s point of leverage: there is no US parent company that could be compelled to disclose. Data control stays with you, not with a foreign provider whose jurisdiction you cannot control.
Can AIgent run air-gapped?
Yes. AIgent can run fully air-gapped, meaning in an environment with no internet connection. On-premise customers install the platform via our Helm chart on their own Kubernetes cluster. It is the same product as the Swiss cloud, just entirely in your environment: neither data nor models leave your infrastructure. Highly sensitive data is processed locally by an on-premise model, and you integrate Entra ID, single sign-on, and role synchronization from your existing directory services. The complete audit trail for regulators and auditors is built into this mode as well. Licensing is per installation, not per seat. Typical settings are the public sector, banks, and healthcare, where data must not leave the premises at all.
Which data leaves Switzerland?
None by default. All data resides in Swiss data centers under revFADP and GDPR, with no transfer to the US. An exception arises only when you deliberately connect an international model such as GPT, Claude, or Gemini for a specific agent, for example for non-critical, non-personal tasks. Even then you control per agent which data reaches which model at all: highly sensitive content stays in Switzerland or in-house, because the corresponding model runs there. The model follows the sensitivity of the data, not the other way around. With on-premise operation, data resides exclusively where you decide, up to air-gapped. Data export and account deletion are built in, so your data remains portable at any time.

Sovereign AI. For every discerning company.

On your use-case, in your environment, within a clearly defined scope. See measurable results within a few weeks.

Secure early access. No per-seat fees. No lock-in.