What is Somia One? A personal AI-native device for the home

Somia™ One is a personal AI-native device developed by yeshcube for access to emotional wellbeing experiences at home through voice and sound. It takes a cubic form, links to an account through Somia App and operates with an internet connection. It integrates the pre-recorded content of Somia Sensory and the interactive experiences of Somia Immersive, with activation through the word ‘Somia’ and complementary control through Somia Ring. This home AI voice and sound device is at the ideation stage: the industrial design, the technical specifications, the price and the commercial release date are determined in the stages that follow.

The ERL scale (Evidence Readiness Level) measures the evidence maturity of a solution across six levels, from the initial hypothesis (ERL-0) to consolidated impact (ERL-5).

The problem: personal home access through voice and sound

Somia experiences are organized around an account, an application and a connected artificial intelligence core. A dedicated home device has to carry that architecture into a personal object that lets a session be completed by speaking and listening, with the application kept for linking and configuration.

Four concrete difficulties follow from that. The first is activating listening in an understandable way and communicating when the microphone is operating. The second is coordinating the device with the account, the preferences, the history and the experiences administered from Somia App. The third is holding the continuity of a session started at another authorized point of access. The fourth is offering physical privacy controls in an architecture that depends on the internet, with no local processing on the device.

The design principle that answers all four is a personal, connected, voice-governed interface, with administration in the application and microphone control on the cube itself.

Design requirements

Five requirements follow from that problem, and they govern the design of Somia One:

• AI-native architecture: artificial intelligence structures the interaction, the personalization and the access to experiences from the conception of the product.
• Voice-led interaction: the word ‘Somia’ starts listening, and sound communicates the responses and the content.
• Personal linking: each unit is associated with an individual account administered through Somia App.
• Explicit privacy: the only physical controls planned manage the microphone, and a subtle visual signal indicates its activation.
• Connectivity and continuity: the device operates with the internet and keeps the authorized session, preferences and progress within the same profile.

Architecture: the cube and the Somia core

Somia’s modular character separates what the core provides from what each product resolves. In that division, Somia One is the physical capture and playback layer: activation-word detection, microphone access, voice and sound output, listening-state indication, physical privacy control, connectivity and account linking.

The core provides the experience orchestration, the content, the safety rules and the data governance, and it coordinates continuity between sessions. Somia App administers the account and the configuration. Somia Ring acts as a complementary external interface.

How a session works

Configuration begins in Somia App. The person links the cube to their account and sets language, connectivity, privacy and preferences. The application administers the device and can start an experience.

In everyday use, Somia One waits for the activation word ‘Somia’. On detecting it, the device opens the microphone and shows the visual indication. The person makes a request and the interaction travels over the internet to the core, which organizes the response according to the mode of use, the safety rules and the authorized data.

The result is played through voice and sound. A session also starts from the application or through Somia Ring. At the end, the history, the preferences and the progress synchronize with consent, so an experience started on another compatible device is resumed when the integration and the permissions allow it.

Modes of experience

Somia Sensory

Somia Sensory brings together pre-recorded sound content for breathing, relaxation, concentration or rest. The sequence and the content are set before playback. On Somia One they are requested by voice, started from the application or activated with a compatible Somia Ring action.

Somia Immersive

Somia Immersive uses the system’s conversational artificial intelligence to adapt the experience to the context and the emotional state the person communicates. A session combines conversation, guided prompts and sound content. The adaptation depends on what is expressed during the interaction and on the preferences the person has authorized. The experience catalog, its durations and the selection criteria will be defined with the product specification.

Data and privacy

The person identifies when the microphone is active and controls which information is linked to their account. The cube plans physical privacy controls and a subtle visual signal indicating microphone activation, and it has no camera. Interactions are processed over the internet, with no local processing on the device, and Somia App centralises permissions, linking and history.

Processing is governed by the GDPR and the Spanish LOPDGDD, with Yeshcube Tech, SL as data controller. The applicable framework is published in the Somia privacy policy and in the AI transparency commitment: personalization memories that are enabled voluntarily, visible, editable and deletable, and an architecture that keeps information on the device by default. What the cube retains on its own account — audio, transcripts, session records and metadata, with their retention period and purpose — will be declared with its technical specification.

Artificial intelligence models make mistakes, read a request incompletely or produce a response that does not suit the context. Oversight of what happens in a session rests with the person.

Evidence and validation status

Somia One holds level ERL-0, initial hypothesis: the purpose, the interaction, the dependence on Somia App, the internet connection and the modes of experience are defined, with no evaluated prototype, sample, test context or measurements of its own.

Level ERL-1 (proof of concept) requires a prototype that allows evaluation of activation-word detection, audio capture and playback, account linking, the privacy controls and the interaction with the core. It will be declared once that evidence is available.

Commercial transfer is conditioned by the hub’s general rule, which requires at least ERL-3 to transfer any solution.

What it is for and what is left out

Somia One is intended for personal use at home and for sessions of breathing, relaxation, concentration, rest and conversation within everyday emotional wellbeing. Where it sits will depend on the industrial design and on the acoustic conditions that are validated.

The product offers non-clinical support. It does not make diagnoses, does not prescribe treatment and does not replace the care of health, psychological or emergency professionals. Left out of its definition are the shared family device, automatic recognition of people, autonomous operation without the internet and local processing.

Somia One takes the place of the personal home interface in the ecosystem, with the application as the administration point and the ring as complementary control. Its compatibility with devices certified under Somia™ Within is conditioned on technical integration. The details of the system behind it are on the Somia site.