Every missed call is a place on the map. Vera answers there.

Cervana AI is a voice AI company. It builds voice agents on its own speech recognition, language and voice models — agents that hold a real conversation and act during it: they book, look things up in your systems, transfer, send e-mails, place outbound calls and write everything down. The same agent runs on a phone line, on a website, inside an app, in a VR environment or a course platform through the API, in 23 languages, in Cervana's EU cloud or entirely on the customer's own servers.

As shipped today: check availability and book, move or cancel appointments (in Cervana's calendar or yours); look up answers in your documents, your website and your systems; act in your systems through MCP and webhooks; transfer to a person by department and opening hours or take a callback; e-mail a code or confirmation; run outbound calls with disclosure and do-not-call handling; spell names and numbers back letter by letter; switch language mid-call; and deliver a transcript, a summary and structured records of every conversation.

No. Phone lines and PBX integration are one deployment. The same agent answers on a website widget and inside apps, and the Cervana API (speech-to-text, conversation, text-to-speech) powers voices in VR learning environments, course platforms and content tools. Customers use it as a receptionist, as a training partner for bank staff, as a follow-up companion between therapy sessions, as a tutor, as a voice for VR characters and, through Cervana Studio, for dubbing and voice generation.

Yes. Connect your systems through MCP and the agent reads and writes them during the conversation — live stock, customer records, bookings. Webhooks deliver every event (call ended, booking made, request recorded) to your software, and your documents and website become its knowledge base.

Plans start at $9.99 a month (Starter: about 83 call minutes and one line). Standard is $229 a month for 1,000 voice minutes and 2 simultaneous calls; Growth ($549) and Scale ($799) add more minutes, more lines and a lower price per minute. Installation on your own servers is quoted per project. All plans include every language, voice and tool.

Yes. Forward your existing number to Cervana, or connect your phone system over SIP. You can also take a new number, or put the voice agent on your website so visitors can talk to it in the browser.

Yes. The agent introduces itself as a digital assistant, which the EU AI Act requires for AI systems that talk to people. Most callers simply appreciate getting an answer straight away instead of waiting on hold.

23 languages: English, German, Arabic, Slovenian, Urdu, Spanish, French, Italian, Portuguese, Dutch, Danish, Swedish, Norwegian, Finnish, Greek, Polish, Russian, Turkish, Hindi, Malay, Japanese, Korean and Chinese. When a caller switches language mid-call, the agent switches with them.

Yes. It books into Cervana's built-in calendar or directly into your own systems through an integration, checks free slots during the call and confirms the time with the caller. Calls, bookings and messages all appear in your console with a full transcript.

A basic receptionist is live the same day: describe your business, add your website or documents, and connect a number. Projects with your own phone system, calendar or CRM integration typically take one to two weeks.

Yes. Cervana's cloud runs in an EU data centre in Frankfurt and call data does not leave the EU. Speech recognition, the language model and the voices are Cervana's own, so no call is passed to a third-party AI company. A data processing agreement (Art. 28 GDPR) is part of the contract, and audio recording is optional.

No. Speech recognition (Cervana Percept), the language model (Cervana Noema) and the voices (Cervana Melo) are Cervana's own models and run on Cervana's or your infrastructure. No call audio or transcript is sent to an external AI service.

On-premise voice AI is voice AI that runs on the customer's own servers — hardware in their data centre or machines in their private cloud — instead of the vendor's cloud. Speech recognition, the language model, the voices and the call logs all run inside the customer's network. Cervana runs on-premise as well as in its EU cloud: the same software, configured for where it runs.

Sovereign voice AI is voice AI where the organisation keeps full control over where its call data is processed and who can access it — typically by running the whole system in its own country or its own data centre, without depending on foreign AI providers. Cervana is sovereign by design because every model in it is its own and it can run entirely on the customer's infrastructure, including air-gapped.

A standard GPU server in your data centre or private cloud. A single server with one 48 GB GPU runs a deployment for one language and several simultaneous calls; more lines and languages scale out with more GPUs. We size the installation to your call volume and languages.

Yes — it is built for organisations that cannot send call data to third parties. Banks, chambers, public bodies and healthcare providers run it on their own infrastructure with no outbound connection to AI services, keep transcripts under their own retention rules, and decide who has access. For UAE and Gulf institutions this is how CBUAE and PDPL data-residency requirements are met.