Our vision: making Europe's companies AI-ready without loss of control

Europe's companies hesitate on AI, and their reasons are rational

In 2025, 20.0 % of EU enterprises with 10 or more employees used AI technologies, up from 13.5 % in 2024. Among small enterprises the share is 17 %. The range across member states runs from 42.0 % in Denmark to 5.2 % in Romania. Source: Eurostat, published on 11 December 2025.

In Germany, 23 % of enterprises with 10 to 49 employees used AI technologies in 2025; the figure was 36 % for those with 50 to 249 employees and 57 % for those with 250 or more (Federal Statistical Office of Germany). The pattern repeats in every statistic: the smaller the company, the rarer the adoption.

The reasons companies give themselves are not hostility to technology but practical obstacles.

We read four brakes into this. The first is cost: a price per workplace falls due before any benefit has been created. The second is missing knowledge, not a missing tool. The third is uncertainty about what is permitted and who is liable in case of doubt. The fourth is a lack of trust in services whose operation and data paths nobody inside the company can inspect.

The cost brake takes effect early and quietly. Anyone paying for every additional person grants access to those who are convinced anyway and leaves the rest of the workforce out. Budgets are tight, and a licence per workplace grows with every new hire. How the vendors cut their licences is set out in the comparison of AI platforms.

The trust brake deserves a clear name: the mistrust is directed at products from outside one's own legal jurisdiction, and it grows with distance. A contract with a vendor on another continent can be read, but not verified. Terms can change, in a dispute a foreign court has jurisdiction, and nobody gets insight into actual operations.

This reticence is a sober risk assessment, not resentment. The same companies work with technology from all over the world as a matter of course; they simply do not want to hand accounting, personnel files and customer data to a service they can neither inspect nor leave. Anyone able to run the system themselves does not need that leap of faith in the first place.

Sovereignty comes from openness, not from assurances

Trust in an AI platform cannot be asserted, trust has to be verifiable. That is why Kisuno will be published entirely under the MIT licence. The basis is LibreChat, an open-source project with more than 40,000 GitHub stars whose source code is publicly available today. The repository with our own extensions will follow shortly.

At Kisuno, open source means that source code, configuration and data paths are open to inspection, including by your own IT department or an external auditor. No feature is locked behind a different licence, and no part of the system runs as a black box.

At Kisuno, openness also means that you can leave at any time. The same software runs on your own hardware, from a single user upwards, with no seat minimum and no mandatory branding. Anyone who hands operations over to us does so out of convenience, not out of technical necessity.

Telemetry is switched off by default in Kisuno, and customer data does not train any models. Anyone self-hosting keeps all content on their own infrastructure. The hosted variant runs in ISO 27001 certified data centres operated by Green in Lupfig, Switzerland; the certification is held by Green, and a dedicated certification for Kisuno is in progress.

We name one limitation openly: web search forwards queries to external search engines, without user identity. The statement that everything stays in Europe applies to storage and inference, not to the search query itself.

What we are building: an AI coworker that stays in Europe

Kisuno is a browser-based AI platform for companies: language models from many providers through a single point of access, processed in Europe, connected to the company's systems instead of being yet another isolated chat window. Files such as Excel, Word and PDF are created in the chat and can be downloaded.

At Kisuno, permissions, costs and traceability are part of the product and not of an add-on package. The connection to the ERP system runs through the MCP gateway, which enforces permissions down to the individual tool: read invoices yes, salaries no. Consumption is visible per user, group and model and can be capped against a shared monthly budget that blocks softly once exhausted instead of quietly running on.

The full feature overview is on the home page, the technical details on running it yourself on the self-hosting page. Access to the hosted Swiss instance is allocated in stages on request because demand is high.

Technology alone does not make a company AI-ready

The most frequently cited reason against AI is a lack of expertise, not a lack of software: 70.9 % of EU enterprises that considered using AI technologies but have not adopted them so far give this reason (Eurostat, 2025 survey). The offering of Herr Informatik GmbH therefore includes training courses and training materials that take employees from the first prompt through to robust analyses.

We train people to verify results rather than trust them. An analysis nobody can recalculate is not progress. Source references, traceable intermediate steps and the clearly stated limits of the tool therefore belong in every training session.

Enabling does not mean replacing. The tasks an AI usefully takes on in a company are routine: summarising, formatting, preparing, looking up. The decision about what happens with the result stays with the employees.

The training materials are created alongside the product and grow with every rollout. We agree scope and format with the company, from an introduction for all employees through to in-depth support for individual teams.

Behind Kisuno stands Herr Informatik GmbH

Kisuno is a product of Herr Informatik GmbH, Klosterzelgstrasse 1a, 5210 Windisch, Switzerland. The company is entered in the commercial register of the Canton of Aargau, UID CHE-112.168.836. Management: Naveein Theivendran and Josip Corković.

Herr Informatik is a Swiss IT service provider and supports companies in their day-to-day operations: infrastructure, operations, support. Kisuno grew out of that work, and so did its requirements. What a business really needs shows up in the support ticket, not in a product presentation.

All details about the provider are set out in the imprint.

Six principles you can hold us to

Six principles apply to the product, documentation and communication of Kisuno: open source, no lock-in, facts instead of adjectives, honest about status, data-minimal by default, traceable. They are deliberately worded so that a breach is noticeable and can be named.

What is in place today and what is not

Kisuno is ready for production use, and we spell out what that means. Available today are requesting access to the hosted Swiss instance, use as an installable PWA, and a configurable retention period including export and deletion of individual user accounts.

• In progress: the GitHub repository with our own extensions under the MIT licence, coming soon, the dedicated ISO 27001 certification of Kisuno, and the general rollout of the Outlook add-in, which currently runs internally in beta, and of the Excel add-in.
• In progress: TOMs document, hosting fact sheet, no-training statement, and an audit and access concept.
• Planned: native apps for iOS and Android, SCIM provisioning, an egress proxy with an allow list for code execution, and hardened sandbox isolation.

What drives us: the potential of European companies

We are driven by what already exists inside European companies. The same knowledge sits there as anywhere else, and the same tasks could be completed faster with AI. What has been missing above all is a tool these companies are willing to entrust with their data. As long as that gap remains open, the productivity gain arises elsewhere.

Using AI on your own terms means the company decides: which data leaves the building, who may see which system, and when the provider is changed. Because the licence does not depend on the number of workplaces, a company can equip all its employees instead of just a selection. Whoever enables everyone gets ideas back from everyone.

Anyone who enables their people while keeping control over data and costs does not have to choose between progress and caution. That is exactly why we are building Kisuno.

Hold us to it. The figures on this page are linked, the six principles above are written so that a breach is noticeable, and if any of it does not add up for you, write to us using the contact details in the imprint. We would rather correct a statement than defend it.

Frequently asked questions