AI Disclosure

Breviya is operated by Nimavera Inc., a Delaware corporation. This page explains how we use artificial intelligence inside Breviya, in line with the transparency obligations in Article 52 of the EU AI Act (Regulation (EU) 2024/1689).

A number of Breviya surfaces are produced by generative AI rather than written by a human author. We list them here so you can identify AI-generated content while you’re using the app:

• AI guided sessions. Personalised meditation, breathwork, and visualisation scripts generated from your stated goals, history, and the technique library. Labelled “AI session” in the player.
• Sleep stories. Narrative bedtime stories generated to a theme and duration you pick. Labelled “AI sleep story” in the library.
• Cinematic Vision Studio sessions. Vision-state scripts tailored to a goal or identity theme you supply. Labelled “AI vision”.
• Daily prompts. Short reflection or check-in prompts surfaced on the dashboard. AI-curated from a pool that mixes human-written prompts and generated ones; generated prompts carry the “AI prompt” tag.
• Breviya Companion. Conversational support inside the practice flow. All Companion responses are AI-generated.
• Programme recommendations. The “Recommended for you” ordering on the programmes and library pages is influenced by AI-derived signals over your stated goals and recent practice history.
• Your “persona” and Mirror. Breviya uses AI to distil a short, evolving profile of what you appear to be working on — themes, goals, and blockers — from your reflections, check-ins, and activity. It quietly personalises the home screen, recommendations, and Companion, and you can view it on the Mirror. This is automated profiling; you can pause it (see “Your controls” below).
• Challenges. Short, proactive multi-day pushes that are selected and framed by AI from your persona.
• Weekly recap & Identity Tape narration. The reflective narrative in your weekly recap, and the monthly “Identity Tape,” are AI-generated from your own activity.
• Voice-to-text. When you record a voice journal or memo, the audio is transcribed to text by a third-party speech-to-text model (you can turn this off in Privacy settings).
• Scene imagery. The illustrated “storyboard” scenes for some wisdom, sleep, and recap content are generated by a third-party image model.

To be clear about the boundary, the following content is authored by humans (Breviya staff or licensed teachers):

• The fixed Silva Method programme curriculum (Foundation, Stress Relief, Goal Attainment) — adapted from José Silva’s original published work.
• The 90-day Identity Reset programme structure, weekly themes, and check-in metrics.
• The Wisdom library entries (stories, parables, and traditions).
• Editorial copy on marketing pages, legal pages, and email templates.

Breviya uses third-party large-language-model providers to power the AI features above. The specific providers and models in use at any given time:

• OpenAI — primary provider for guided sessions, sleep stories, vision scripts, and Companion responses. Current default model: GPT-4o-mini.
• OpenRouter — a routing layer used as a fallback for text generation and, by default, for voice-to-text transcription (a Whisper model, currently hosted via Groq). The underlying model varies and is logged per-session in our internal audit trail.
• Replicate — third-party host for the image-generationmodel that renders illustrated storyboard scenes. Only the text prompt for a scene is sent; your journal or reflection content is not.
• Kokoro TTS — open-source text-to-speech model run on Breviya’s own infrastructure (no third-party audio provider) for sleep-story, guided-session, and recap narration, in English and Spanish. Your narration audio is generated server-side and is not sent to any third party.

Model defaults can be updated by Breviya operators through the admin console. We will update this page whenever the primary providers change. The full list of third-party providers that process your data is on the Sub-processors page.

Every AI-generated surface in Breviya carries one of the following identifiers:

• A visible label or badge — “AI session”, “AI sleep story”, “AI vision”, “AI prompt” — on the listing card and on the playback / detail view.
• A “Generated for you” attribution line in the session metadata, including the date the content was generated.
• For Companion exchanges, the conversational UI is itself labelled “Breviya Companion” with a notice that responses are AI-generated.

If you encounter any AI-generated content inside Breviya that is not labelled as such, please tell us using the channel below — we treat unlabelled AI content as a transparency bug.

AI features are core to Breviya’s product, so there is no app-wide toggle to disable them. You can, however, do the following:

• Use only the human-authored content — Silva, the 90-day Identity Reset, the Wisdom library, and the breathwork lab — without triggering any AI generation.
• Pause your persona. In Settings → Privacy you can pause personalization so the AI stops distilling and updating your persona. Your existing persona is left in place, read-only, until you turn it back on.
• Turn off voice transcription. In Settings → Privacy you can stop voice recordings from being sent to the third-party speech-to-text provider; you can still type your entries.
• Keep journals out of personalization. Reading your journal content to personalise the Companion and persona is off by default and opt-in; you can withdraw it at any time in Settings → Privacy.
• Request export or deletion of any AI-generated content tied to your account via Settings → Privacy → Export / Delete. Your AI-generated sessions are treated as personal data and are covered by the same GDPR rights as the rest of your profile.

Generative AI can produce text that is inaccurate, awkward, or off-tone. Breviya mitigates this through prompt engineering, light moderation, and human-authored fallbacks, but cannot fully eliminate it. In particular:

• Breviya is not a medical service. AI-generated content does not constitute medical, psychological, or therapeutic advice. See the Medical Disclaimer in the Terms of Service.
• Companion is not a crisis line. To respond safely, Breviya screens Companion messages for self-harm or crisis language (using a moderation model plus a local keyword check) and, when triggered, replies with crisis resources instead of a normal recommendation. This screening only surfaces resources to you — it does not notify any third party, emergency service, or authority. If you are in distress or considering self-harm, please contact your local emergency services or a crisis line; Companion cannot replace human help.
• The persona is not an assessment. The AI-distilled persona and its recommendations are a convenience for personalisation, not a clinical or psychological evaluation of you.
• AI-generated sleep stories and visualisations are entertainment / wellness content, not diagnostic or therapeutic outputs.

If you have a concern about AI content you received in Breviya — unlabelled AI content, harmful output, copyright or privacy concerns, or any other transparency issue — please email ai@breviya.com. We commit to acknowledging within 5 working days and resolving substantive issues within 30 days.

Under Article 85 of the EU AI Act you also have the right to lodge a complaint with the AI office or competent national authority in your member state.

We will update this page when the AI providers we use, the surfaces that use them, or the controls available to you change. The “Last updated” date at the top reflects the most recent material change.