Deeptech and AI start-ups
A high-performing model, a strong research team, and the question of adoption by business customers. We work on interaction, evaluation and trust.
AI consultant
in Lausanne, from prototype to real use.
As an AI consultant for start-ups and companies in Lausanne, we help with the step that even the best models cannot take on their own: turning them into a tool people actually use. Human-AI system design, evaluation protocols and team adoption are at the core of our work in Vaud.
After the model, the use.
In the wake of EPFL and UNIL, the Vaud ecosystem gives rise to many young tech companies, a good number of them with a strong machine learning culture. Their difficulty rarely lies in the algorithm. It shows up when a convincing prototype has to become a feature that a business customer understands, adopts and trusts for good reason.
That is where we step in as an AI consultant in Lausanne. Together with the team, we define when the user stays in control, how to display how certain an answer is and which metrics will prove the feature really helps. These design choices often weigh more heavily on adoption than a few performance points gained on a test set.
The canton of Vaud is not just start-ups. Its manufacturing SMEs, service firms, and hospitality and tourism businesses along the Riviera are asking questions too. For them the stakes are more down to earth: lightening the admin load, responding to enquiries faster, making better use of documents built up over years, all in line with the Swiss FADP.
For teams across Vaud.
Those building AI as much as those who want to use it.
SaaS vendors and scale-ups
A growing product and an AI feature to add without weakening the existing experience: use case choice, mock-ups, measurement protocol, integration.
SMEs in the canton of Vaud
Manufacturing, services, retail: a few concrete generative AI uses, simple governance and trained staff, in line with the Swiss FADP.
Hospitality and tourism on the Riviera
Replies to enquiries, multilingual content, guest reviews: AI prepares, your teams keep the human touch your reputation rests on.
Our work across Vaud.
For teams building AI as much as for those who want to use it.
Human-AI system design
Who decides, when, with what information: we design the interaction between your model and its users, right down to mock-ups of the screens involved.
Evaluation protocol
Representative test cases, quality criteria and observation in real conditions, defined before development so you know whether the feature delivers.
Diagnostic for Vaud SMEs
A review of your processes and data, then a shortlist of realistic uses, each assessed for time saved, effort and risk.
Upskilling
Hands-on workshops in Lausanne or on the Riviera for business teams, with a usage policy designed around Swiss data protection requirements.
What you receive in Lausanne.
Deliverables designed for a product team that wants to decide quickly and well.
- Human-AI system specification: who decides, when, with what information.
- Mock-ups of the screens involved, including uncertainty and error states.
- A set of representative cases and quality thresholds to reach before launch.
- A plan for measuring adoption once the feature is live.
- For SMEs: a shortlist of assessed uses and a policy that meets your requirements.
- Materials from workshops held in Lausanne or on the Riviera.
What your clients search on Google.
- AI consultant Lausanne
- AI strategy consulting Lausanne
- integrating a machine learning model into a product
- AI training for teams Vaud
- AI for SMEs canton of Vaud
- how to evaluate an AI feature
People keep the final say.
A strong model is not enough: you need a system where people understand and decide. At Metroscope, the AI anticipated failures, but EDF’s nuclear engineers remained wary. Victor Michel, the studio’s founder, observed their work in the plants, designed alerts graded by criticality and supported the rollout with the teams. That human-AI design is what we bring to Lausanne teams.
EDF · Metroscope · Nuclear energy · Industrial AI · 2018–2019
- 56
- nuclear reactors operated in France
- IA
- industrial predictive maintenance
What to know in Lausanne.
What we see when an AI prototype has to become a product.
Show the uncertainty
A model is never one hundred per cent sure, and the interface should say so. Showing a confidence level, flagging edge cases, prompting users to check when an answer is shaky: these choices stop users following the machine at the very moment it is wrong. That is often the difference between a feature people put up with and one customers trust for good reason.
Usage metrics, not just performance
A machine learning team measures accuracy on a test set. The product must also measure usage: how many suggestions are accepted, edited or ignored, how much time users save, whether they come back. These indicators are defined before going live; they reveal problems the test set cannot show, such as a correct suggestion in the wrong place, or one that arrives too late to be useful.
Criteria for choosing a product studio
Before entrusting the design of an AI feature, check four things. Has the studio shipped products used in production, not just mock-ups? Does it work with your engineers rather than beside them? Does it propose an evaluation protocol before designing? Can it say no to a feature that adds nothing? Finally, ask who actually does the design: those answers say more than a portfolio.
The lake is no barrier to method
Working with a French studio from Lausanne mainly raises practical questions: data protection, contracts, travel. The Swiss FADP is built into our recommendations from the scoping stage, workshops take place at your offices and work moves forward online between visits. The founder remains your single point of contact, from first conversation to final debrief, with no intermediary or sales go-between.
Where we work.
Only Lake Geneva lies between Thonon and Lausanne: design workshops and training sessions take place at your offices, and work carries on online between visits.
- Lausanne
- Renens
- Pully
- Morges
- Vevey
- Montreux
- Yverdon-les-Bains
Frequently asked questions.
AI consultant Lausanne: answers to the questions we are asked.
Our start-up already has data scientists: what does an AI consultant add?
A product and usage perspective. Your engineers build the model; we work on everything around it: its place in the user journey, human oversight, how uncertainty is shown, how adoption is measured. The two skill sets complement each other without overlapping.
How do we know an AI feature is ready for our customers?
By defining in advance what “ready” means: a set of representative cases, acceptable quality thresholds and a test with real users. Without that protocol, a successful demo can hide inadequate quality once the feature is live.
Do you also work with SMEs in the canton of Vaud?
Yes, just as we do with start-ups. For an SME, the engagement focuses on a few concrete uses, simple governance and training for the staff concerned, with the Swiss FADP taken into account.
Can the workshops be held in Lausanne?
Yes, at your offices in Lausanne, Renens or on the Riviera. Design sessions work best in person; follow-up and interim reviews take place online.
What should you check before choosing a product studio?
Products shipped and in use, a written method, the ability to work with your engineering team and an evaluation protocol before development. Ask who will do the work too: at Brutux Studio, the founder designs, from the first call to delivery.
Can you integrate the feature into our product?
Yes. We can stop at design and evaluation, or go all the way to integration in your codebase, alongside your developers. The choice depends on your team’s strengths and your timeline.
Does our Swiss customers’ data leave the country?
Not without an explicit decision on your part. We compare the providers and hosting options available, in Switzerland and elsewhere, and set out their implications under the Swiss FADP before any choice is made.
How long does an engagement in Lausanne take?
An AI consulting engagement takes 4–8 weeks. For a start-up, it ends with a feature that is specified, mocked up and equipped with its evaluation protocol; for an SME, with prioritised uses and trained staff.
(Let’s work together)
Let’s talk about your project.
30 minutes to talk about your project, in Thonon or by video call. I’ll tell you honestly what will make the difference.
Book 30 minutes ↗