Your privacy, in plain English

Leyra is designed around a simple promise: remembered guidance should remain under your control. This notice explains what Leyra collects, why it is used and how to ask for help with your data.

Leyra.uk is operated from the UK. For privacy questions, data requests, account deletion requests or concerns, contact support@leyra.uk.

This notice is written for the current launch service and will be updated when legal entity, vendor or infrastructure details change.

• Account details: email address, display name, encrypted password record, confirmation state and account settings.
• Birth and profile details: birth date for account readings, and optional birth time/place, tone preference, focus area and reflective/profile answers.
• Reading and ritual history: saved readings, moon/tarot activity, journal/check-in signals, profile-memory states and related preferences.
• Subscription details: Leyra plan, Stripe customer/subscription references, subscription status and billing-portal state. Leyra does not store full card details.
• Email preferences and delivery signals: whether ritual email is enabled, delivery mode, frequency, email kind, optional weekly reflection email day and preference changes.
• Feedback and support context: message, category, optional contact email if supplied, page path, browser/device context and account tier when signed in.
• First-party analytics: only after optional analytics consent, Leyra records first-party page/event data such as path, page title, referrer without query strings, viewport, language, timezone, anonymous session ID and event labels.
• Public share-card signals: Leyra may record privacy-safe card views, channel clicks, copy/download events and free-reading/pricing CTAs after analytics consent. These events use public paths and labels, not private reading text or account memory.
• Home Screen install and offline app signals: Leyra links app-install metadata, may remember if you dismiss the install suggestion in that browser, and registers a service worker that caches only public static shell assets plus a safe offline page. Push subscription device reminder details is collected only when a signed-in user explicitly enables browser reminders on that device.
• Safety and quality records: AI safety-review rows can include prompt/output excerpts so unsafe or low-confidence readings can be investigated and improved.

Leyra uses data to provide accounts, confirm email addresses, generate sign-based and personalised readings, maintain account memory and conversation continuity, manage subscriptions, deliver selected emails, keep the service secure, answer support requests, improve usability during launch and maintain safety guardrails.

The usual lawful bases are contract where data is needed to provide the service you request, legitimate interests for security, support, safety and product-quality work, consent for optional analytics storage, and legal obligation where records must be kept for accounting, tax, dispute or regulatory reasons.

The installable web app layer lets Leyra be added to a phone Home Screen using browser support. Leyra now registers a conservative service worker so static shell assets and a no-private-data offline page can load more reliably. Personal readings, account pages, billing pages, profile memory, saved history, tarot results, analytics/API responses and private admin routes are excluded from the cache by design.

Leyra supports an explicit browser push opt-in for signed-in users. When you enable reminders, Leyra stores the the device reminder details needed to manage that reminder so that reminders can be managed safely. Leyra does not use push for third-party advertising and does not include private reading text in notifications. You can disable the current browser subscription from your profile delivery settings or your browser/device notification settings. Reminder preferences also store morning/evening channels, quiet hours and delivery records before scheduled pushes are enabled.

Daily, moon and tarot share cards are public growth surfaces. They are generated from public date/sign/moon/tarot copy only and must not include private reading text, journal answers, profile-memory answers, paid tarot history, email addresses, confirmation tokens or billing context.

Weekly reflection summaries and optional weekly reflection emails use aggregate theme signals such as repeating theme, one thing that helped and one question to carry. The email copy avoids raw private journal answers and can be paused from profile delivery settings.

Leyra uses AI-assisted writing inside safety rules. Sensitive account-memory prompts are handled inside Leyra’s approved private AI setup unless a future policy clearly says otherwise. Leyra aims to minimise prompt context and avoid sending more profile memory than is needed for the reading tier and feature being used. Personal tarot keeps card selection deterministic, but may use the approved Leyra AI setup to shape the interpretation text from the chosen card, orientation and relevant profile context.

Leyra may also use infrastructure, email-delivery and payment vendors. Stripe handles checkout, customer billing and card details. Leyra keeps subscription references and status, not full card numbers.

Account, subscription and reading records are retained while the account is active and for a reasonable period afterwards where needed for support, safety, billing, fraud prevention or legal obligations. Launch analytics and feedback are intended for short operational review rather than permanent profiling. Self-service export, profile-memory deletion, account deletion/deletion-request handling, anonymisation and retention operations are available from signed-in privacy controls.

You can edit email preferences, profile basics and optional profile-memory material inside your account. Signed-in users can also open Privacy controls to download a data export, delete profile memory, start account deletion and see what account-linked records exist. Leyra also publishes a public account deletion and Meta app data deletion instructions page. You can also contact support to request access, correction, deletion, export or restriction of your data.

UK users can raise unresolved data-protection concerns with the Information Commissioner’s Office. Please contact Leyra first where possible so the issue can be investigated quickly.

Leyra is intended for adults aged 18 or over. Adult-only features require a valid saved date of birth. Accounts that should not be using the service may be restricted or deleted.

Leyra readings are for reflection, wellbeing and entertainment. They are not medical, legal, financial, emergency, therapy or professional advice. In an emergency or crisis, contact local emergency services or an appropriate crisis-support organisation.

Memory starts active for newly created registered accounts so Leyra can provide continuity without requiring people to discover a hidden switch first. Existing saved choices are respected: an account that has paused memory remains paused, and a channel-specific opt-out remains off.

Leyra may retain eligible signed-in web, verified-email and linked-WhatsApp conversation records, profile preferences, reading tone, preferred length, spiritual/practical balance, current focus, people or pets you explicitly mention, projects, goals, routines, boundaries, check-ins, feedback and longer-view conversation summaries.

Not every stored sentence becomes a permanent fact and not every remembered detail is sent to every response. Leyra separates source conversation history, durable memory, details waiting for confirmation and rolling/monthly/yearly summaries. At response time it selects only relevant, eligible context. Current information takes priority over superseded or expired information.

Potentially sensitive health, relationship or financial details require additional confirmation before becoming active durable memory. Passwords, access tokens, private keys, payment-card details and one-time codes are excluded from durable memory and normal context selection.

Memory data is used to provide Leyra's account and companion features inside the approved private AI setup; it is not used to train shared AI models. You can use What Leyra remembers or My data & Leyra memory to inspect current and historical information, preview what may be selected, pause memory, pin or remove items, export your data or delete profile memory.

You may also give a clear instruction to remember, keep important, change or forget one detail. Leyra records a limited outcome audit so the instruction is not applied twice, without creating another unrestricted copy of the private instruction text. Vague changes are not applied, sensitive details still require confirmation, and security credentials are rejected.

Leyra daily check-ins may store mood, energy, need, an optional short note, the remember-mode choice, channel, safety-scan result and timestamps. The check-in can shape the day’s reading immediately.

If you choose “use for today only”, Leyra should not promote that check-in into future profile memory or active themes. If you choose “remember as a pattern” and memory is active, Leyra may use the short summary as a gentle pattern signal. Optional notes are scanned before storage and may be suppressed if they match high-risk safety rules.

Leyra may store a short reading-support record when a reading is generated. This records which consented sources were used, such as profile settings, today’s check-in, recent reading memory, tarot memory or structured feedback signals, plus a private reference for the reading support record. It is designed to support personalisation and safety without keeping long raw histories in admin screens.

When you rate a reading as helpful, vague, too intense, not right today or explicitly ask Leyra to remember something, Leyra may store that structured feedback so future readings can adapt within your memory-consent setting. You can export these rows from Privacy controls, and deleting profile memory removes the feedback-learning and reading-support records linked to your account.

Leyra may suggest short active themes from remembered check-ins and reading feedback when memory consent is active. These are user-visible, confidence-scored and removable; they are used only to make Leyra readings feel more personal and are not shared model training data.

Signed-in users can review the profile preferences, boundaries, daily check-ins, feedback signals, active themes, pending memory candidates and context-use audit rows Leyra may use for personalisation. Memory can be paused or resumed, and data export/delete controls remain available from the same account area.

Ask Leyra stores the question, generated answer, safety decision and reading-support details so you can review private reflections and so Leyra can audit safety. While memory is active, clear normal-sensitivity facts and small meaningful details may become source-linked memory; uncertain or sensitive details remain waiting for confirmation. When memory is paused, stored history remains available to you but is withheld from normal historical context selection.

Before live WhatsApp use, Leyra treats WhatsApp as a gated inbound Ask Leyra channel. Inbound test messages may be stored with hashed phone/WhatsApp identifiers, message text, command status, safety-review scores and admin-only response previews. When a confirmed phone number links the contact to a registered account, new linked contacts inherit that account's active memory setting unless the contact has opted out or requested memory deletion. STOP and DELETE MEMORY disable WhatsApp-derived memory contribution.

When WhatsApp testing is enabled, Leyra may store private request records and admin preview decisions so STOP, START, HELP, STATUS, PRIVACY and DELETE MEMORY requests can be checked before any live pilot. These records are private account or contact records and are not used for public share cards.

If you join the WhatsApp pilot waitlist, Leyra stores the email address you submit and the page context needed to process the request inside the feedback/admin inbox. This does not enable live WhatsApp replies, and it does not give WhatsApp-derived memory permission.