gensai.aiGensai designs custom voice AI solutions: voice agents that answer the phone, voicebots on kiosks or mobile, callbots for reception and appointment booking, AI avatars that give the voice a face. Connected to your data through RAG, they answer with controlled latency, in the visitor's language, hosted in France or in Europe, or fully on site.
An AI voice agent chains three building blocks around a language model connected to your data: speech recognition, reasoning, speech synthesis. Everything comes down to latency and how well the dialogue holds. With real-time APIs or models running on site, latency is kept to a minimum, on the phone as on a kiosk, and the conversation stays natural, with its interruptions and resumptions.
On the phone, the callbot picks up, understands the request, informs, qualifies and books the appointment in the calendar, including outside opening hours. On site, the voicebot welcomes visitors, answers practical questions or tells the story of an exhibition. The AI avatar, a 3D character synchronised with the voice, turns the assistant into a real interlocutor, in a museum, at a trade show or at the entrance of a venue. To try it, Gensai's own avatar answers live: just talk to it.
And when confidentiality or the venue requires it, everything runs fully on site, with no Internet connection, no cloud, no tokens: Gensai deployed a complete voice installation on a simple Mac Mini for a busy aquarium. A robust, energy-frugal choice, precious for scenography as well as for sensitive data.
Use cases for voice agents and voicebots
Intelligent phone reception
A callbot qualifies requests, informs and takes messages, even outside opening hours, then hands over to a person when needed.
Appointment booking by phone
A voice agent picks up, understands the request, checks the calendar and confirms the appointment, in real time.
Cultural mediation in museums
A conversational museography installation: the visitor talks with an artwork, a character or an exhibition, by voice or on screen.
Tourist office reception
A multilingual voice assistant available around the clock: practical information, visit ideas, personalised recommendations.
Leisure parks and attractions
Waiting times, park map, food, ticketing: a voicebot answers families, on a kiosk or on mobile.
Welcome avatar
A 3D character embodies your brand at the entrance of a venue, on a stand or on your website, and starts the conversation.
Automated oral assessment
Run language or skills assessments orally, at scale, with consistent scoring.
Phone call analysis
Transcribe and analyse thousands of calls: pledges, quality control, compliance, customer verbatims.
Our references
Frequently asked questions
What is an AI voice agent?
An AI voice agent understands speech, reasons with a language model connected to your data and answers out loud. It combines speech recognition, language understanding, RAG and speech synthesis. Unlike menu-based voice servers, it holds a natural dialogue and answers open questions.
What is the difference between a voicebot, a callbot and a voice agent?
All three share the same technical core. The callbot is the voice agent on the phone (reception, qualification, appointments). The voicebot is the voice agent on a device, kiosk, application or website. Voice agent is the generic term, and the voice assistant its consumer cousin.
Can a voice agent work without the Internet?
Yes. Gensai has deployed complete voice installations running on site on a simple Mac Mini: speech recognition, language model, RAG and speech synthesis, without any connection. An ideal approach for museography and busy venues.
What latency does a natural dialogue need?
The shorter the delay between the end of the question and the start of the answer, the more natural the dialogue feels. When it stretches, the user repeats or hangs up. Gensai measures and optimises this latency on every installation, with real-time APIs or models running on site, and tunes the agent's speaking pace.
What is an AI avatar?
An AI avatar is an animated 3D character that embodies a conversational assistant: it speaks with a synthetic voice, stays in sync with it and reacts to the exchange. It gives your brand a face, on a website, a stand or a welcome screen.
Can a callbot replace a phone switchboard?
It complements it. The callbot absorbs recurring calls (opening hours, access, appointment booking, first-level information), at any time, and hands over to a person as soon as the request leaves its scope or the caller asks for it.
Which languages can a voice agent answer in?
Most common languages: speech recognition and synthesis are multilingual, and the agent can detect the speaker's language to answer in the same one. Handy for a tourist office, a museum or an international trade show.
Where is voice data hosted?
With French or European hosting providers, and entirely in France if you ask for it. For sensitive venues or places without a connection, everything can run on site, without any voice leaving the device.
A voice AI project?
Describe the journey of your visitors or your callers: Gensai designs the voice agent that answers them, from the first prototype to production.