Saltar al contenido
NNexunia
legal

How we use AI

In force since 14 October 2026Version 1.2

Courtesy translation: the official version of this information is the Spanish one.

When you talk to a business that uses Nexunia — on WhatsApp, on its website, by email or by phone — you may be getting a reply from an artificial intelligence assistant rather than a person. This page explains, in plain language, what that assistant does, what information it uses and what always stays outside its scope.

Understand what you ask for

The language model reads the conversation (WhatsApp message, widget chat, email or call) and works out what you need: information, a booking, a change or to speak to a person.

Look up real business information

Prices and availability of time slots always come from the business's real tables, queried at that moment — the model neither generates them nor remembers them from a previous conversation. FAQ-type questions are answered against the documentation that the business itself has uploaded, never against the model's general knowledge about that business.

Carry out actions only through verified tools

The model never runs code, queries the database directly or receives third-party credentials. When it needs to create, change or cancel an appointment, it calls a specific tool that validates the request against the real business and returns a verified result — an appointment only exists once that tool confirms a real identifier, never before.

Summarise and translate for the business

The language model also summarises calls handled by AI and, when the business requests it from its panel, translates into Spanish the transcript of a conversation (voice or chat) written in another language, so that its team can understand it. The original transcript is always kept intact and the translation is identified as automatic.

Hand over control to a person

At any point in the conversation, the customer can ask to speak to a person at the business, and a team member can take over the conversation from their panel — on phone calls, this may involve transferring the call to a human number if the business has configured it.

What falls outside the scope of the AI

It does not replace regulated professional advice

The assistant does not give medical diagnoses, legal or financial advice, or any other recommendation that requires a specific professional assessment, unless the customer business has expressly configured and validated that type of content for its own case.

It does not make decisions outside what is configured

The assistant acts within the tools and policies that each business switches on for its account (for example, what details to ask for before booking, or whether to escalate a complaint). It does not decide on its own to widen that scope.

Recording and transcription of voice calls

When the conversation is a phone call handled by AI, the text is always transcribed, just as on the other channels. If the business has call recording switched on, the audio of the conversation is also recorded — so it can be reviewed in the panel and so that a person who takes over has real context of what was said — and this is announced at the start of the call; without that feature switched on, the call is not recorded as audio. This applies equally to Nexunia's public demo calls (the “Call me now” form) and to manual calls that a human agent makes from the panel.

Which providers process the conversation

The language model that interprets the conversation is one of the leading commercial models available today, chosen for its capability and its security guarantees. On phone calls, the main voice engine is a real-time voice model that receives the call audio, transcribes it, generates the reply and returns it spoken, all in a single session — in that case, the audio and the content of the conversation are processed by that voice provider, while actions on the business (bookings, changes, cancellations) are always executed by Nexunia's backend. Depending on the business's configuration, some accounts use an alternative voice engine, with separate speech recognition and synthesis and additional providers. The summary by processing category can be found under providers that process data. A customer who has already signed up finds the named list of providers in their panel; anyone else can request it at privacy@nexunia.com.

This page does not replace the notice you may receive at the moment of the conversation itself (for example, at the start of a call) — the two complement each other. Questions about a specific case? Write to us.