A Lausanne-based startup has announced two artificial intelligence models designed to bypass American infrastructure. According to Netzwoche, Giotto is launching Solo and Alto, two logical reasoning tools that explicitly target "digital sovereignty." For a Romandy SME owner, this raises a straightforward question: is this a real opportunity, or a specialist debate with no bearing on day-to-day operations?
The answer depends on what data your business handles, who bears technical responsibility, and what you actually expect from AI on a daily basis.
What Giotto Is Actually Announcing
Giotto, based in Lausanne, develops AI models specialized in logical reasoning. Its approach relies on Test-Time-Compute: rather than training ever-larger models, the company allocates more computing resources at the moment of use. This method aims to reduce infrastructure requirements while maintaining performance.
Two models are on offer. Solo runs on a single graphics card (GPU), making it theoretically deployable on a standard workstation. Alto, more powerful, requires eight Nvidia B300 GPUs and targets intensive professional use cases. Their distinctive feature lies in the hosting options: shared cloud in Europe, dedicated cloud in Switzerland or elsewhere in Europe, or direct installation on the company's own hardware.
CEO Aldo Podestà emphasizes independence from American infrastructure. Giotto's cloud services claim GDPR compliance. An anonymous chat service without user accounts, mobile applications, and a developer API round out the launch offering.
What Changes for a Romandy SME
The promise of local or European hosting addresses a genuine concern. The Swiss nFADP (nLPD) and European GDPR impose guarantees on the processing of personal data. For a Valais-based accounting firm analyzing payslips, a Vaud medical practice drafting reports, or a Neuchâtel hotel centralizing guest preferences, knowing where information transits matters.
However, technical sovereignty is not declared by fiat. Consider a carpentry business in Sierre wanting to use Solo locally. It would need to acquire a machine with a compatible GPU, install it, maintain it, secure it, and ensure a reliable connection. The time cost for an employee or external IT specialist typically exceeds any savings on API requests.
| SME Profile | Viable Option | Likely Obstacle |
|---|---|---|
| Retail business (5-10 employees) | Shared European cloud | Integration with POS and accounting |
| Accounting firm (15 employees) | Dedicated Swiss cloud | Validation with data protection officer |
| Property management company | Local if large document volume | Hardware and maintenance costs |
| Engineering consultancy | API for specific internal tool | Development skills required |
The Alto model, with its eight GPUs, remains economically out of reach for the vast majority of Romandy SMEs. It targets organizations that already have substantial IT infrastructure or service providers.
Hypothetical Scenario: A Property Management Company in Martigny
A typical case: a property manager oversees 400 units. Each week, dozens of emails arrive regarding breakdowns, regulations questions, transfer requests. One employee spends roughly six to eight hours on this.
The company considers an AI assistant to pre-classify these requests and draft response templates. With an Atlantic-hosted model, the board expresses reservations about tenant data. The Giotto option in Swiss or European cloud could reassure, provided the contract specifies the actual server location and any subcontractors.
Even on European cloud, the company would need to verify three points: the ability to recover its data if switching providers, the procedure in case of a security incident, and the availability of French-speaking technical support. These elements do not always appear in a product's launch documentation.
Points to Watch Before Moving Forward
Hosting and the nFADP (nLPD). nFADP compliance does not automatically follow from being "in Switzerland" or "in Europe." It requires a processing agreement, documented technical and organizational guarantees, and often a data protection impact assessment. For specific questions, consult the Federal Data Protection and Information Commissioner (FDPIC) or a qualified specialist.
Real costs. The free Alto on shared cloud attracts attention. In a hypothetical scenario, an SME might sign up, then find its needs exceed the free limits. Migrating to a paid plan, or even to another provider, consumes internal time. The API is typically usage-based billing.
Product maturity. Giotto was founded recently. Its models have not yet seen wide adoption. For a Romandy SME, monitoring developments for several months before committing a critical business process remains prudent.
Technical dependency. "Europe must offer more than just a different flag on the same dependency," Podestà told Bilan. This quote reflects an ambition. For the user, dependency does not disappear: it shifts provider. Moving from Microsoft to Giotto, or from OpenAI to a local model, is still an outsourcing relationship.
What to Do Now
- Inventory your sensitive data. What information, if leaked, would cause problems for your customers, employees, or reputation? This classification then guides your hosting choice.
- Test without critical commitment. Giotto's free anonymous chat allows a first evaluation of logical reasoning quality. Do not yet integrate it into a business process: simply compare response quality with your current tools.
- Check your existing contracts. Your office suite, CRM, or accounting software may already embed AI whose hosting location you do not know. This is often where the real risk lies.
- Demand proof of location. If a provider promises "Swiss cloud," require the datacenter address and associated certifications. A generic claim does not suffice to justify due diligence in case of an audit.
- Evaluate total cost of ownership. Include any hardware, maintenance, training, and supervision time. A well-integrated American cloud solution sometimes costs less than a poorly managed "sovereign" alternative.
At FlowBiz.ai, we build assistants and automations for Romandy SMEs, with particular attention to data location and gradual deployment. If you want to explore what AI could bring to your business without committing your infrastructure from day one, our free selection assistant and pre-diagnosis tool identifies use cases suited to your situation. You can also review concrete automation examples or take our two-minute AI quiz.



