Swiss startup Oravita brings unscripted AI conversations in virtual reality to Slush 2026

A decade after the idea first took shape in a Swiss language school, Oravita will show a working VR platform where learners speak freely with an AI native speaker inside realistic everyday scenes.

ZUG, Switzerland, 2 October 2026. Oravita, a Swiss startup that offers an add-on to language classes to help learners improve their speaking skills in virtual reality, will present its platform at Slush 2026 in Helsinki on 18 and 19 November. At the Swiss Pavilion, visitors can put on a headset, explore the virtual city and talk with an AI avatar who replies in real time. There is no script to follow. The learner speaks, and the AI avatar replies.

An idea that waited for its technology

Oravita began with a request that Isabelle von Niederhäusern heard again and again. When she opened her language school ten years ago, her clients kept asking for one thing: more chances to speak. Grammar and vocabulary can be studied alone. Conversation needs someone on the other side, and in a group lesson each learner gets only a few minutes of it.

“Our learners understood a great deal, but when a real conversation started, many of them froze,” says von Niederhäusern, co-founder of Oravita. “They did not need more exercises. They needed somewhere to practise speaking, as often as they wanted, without feeling judged.”

The answer came from gaming. Video games had shown how quickly people feel at home in a virtual world and how naturally they move through it. Von Niederhäusern began to imagine a world built for language learning, where every object, shop sign and passer-by is something to talk about.

For years, the hardware and the speech technology were not good enough to make that work. She kept the idea alive through calls with researchers, engineers and entrepreneurs, and waited until headsets, speech recognition and language models had caught up.

Selected by Innosuisse, developed with ETH Zürich and professional partners

The search for professional partners led to an innovation project funded by Innosuisse, the Swiss Innovation Agency. In Innosuisse’s 2025 call for SME innovation projects, Oravita was one of 81 projects selected from around 270 applications, with ETH Zürich as its research partner.

Since January 2026, a team of researchers, LLM and AI avatar specialists, VR world developers and language teachers has turned the concept into a working prototype, currently in two languages, English and French. More languages will follow in 2027. It was built to professional, industry standards, and the research behind it has been published by IEEE.

Oravita has opened a waitlist for its beta. Individual learners, language schools and companies have already signed up to be among the first to try it.

“Ten years ago this was a sketch on paper,” says von Niederhäusern. “Today you can put on a headset and speak about any topic in another language, and the AI avatar speaks to you as a native speaker, in real time. That still surprises me every time.”

Built for the way people really speak

The next stage of development focuses on something most language tools leave out: the same language sounds different from one country to the next. Portuguese in Lisbon is not Portuguese in São Paulo, and Spanish in Madrid differs from Spanish in Mexico City or Buenos Aires. Oravita is building its platform to reflect these differences in pronunciation, vocabulary and everyday expressions. According to the company, it is the only VR language platform in the world to account for dialects in this way.

The people learners meet in Oravita are designed as native speakers of the place they represent. How they greet, how formal they are and how they react follows the culture of that country. Learners pick up how people behave in a conversation as well as what they say.

Every conversation happens in real time and none of it is scripted. The AI avatar also knows what surrounds the learner. A learner can point at a menu, a ticket machine or a pastry in a shop window, and the avatar will give them the right word and the grammar to go with it.

Oravita is also in talks with brands and tourism organisations about scenes based on real places. A traveller could walk through a square in the city they are about to visit, or order at a café they will later find on the corner. Oravita will announce partners once agreements are in place.

“We want people to arrive somewhere new and feel that they have been there before,” says Marc Marchon, co-founder of Oravita. “Language opens that door, and culture keeps it open. A conversation in Lisbon should feel like Lisbon.”

Meet Oravita at Slush 2026

• Dates: 18 and 19 November 2026

• Venue: Helsinki Expo and Convention Center, Messuaukio 1, 00520 Helsinki, Finland

• Stand: Swiss Pavilion (swisstech), organised by Switzerland Global Enterprise and Swiss Business Hub Nordics, stand 7A11

• Live demo: Visitors and journalists can try the prototype at the stand. Interviews with the founders can be arranged in advance.

About Oravita

Oravita is a product and brand of abc multilingual collective AG, based in Zug, Switzerland. Its platform combines virtual reality with conversational AI so that learners can practise speaking a language in realistic everyday scenes. It is an add-on to language classes. Learners talk with an AI avatar in real time, without scripts, at their level on the Common European Framework of Reference for Languages (CEFR), from A1 to C2. Oravita works with individual learners, language schools, universities and companies. Its research is supported by Innosuisse, the Swiss Innovation Agency, with ETH Zürich as research partner, and its prototype has been published by IEEE. Oravita is currently in beta. More at oravitavr.com.

Media contact

Hannah Schlösser, Head of Sales, Oravita
oravitavr.com/contact · Request a call

Images, founder portraits and the Oravita logo are available at oravitavr.com/downloads.