Privacy Policy

1. Who is responsible

Lila is operated by Furkan Efe Genç, a sole proprietorship registered in Türkiye, who is the data controller for the information described here.

Registered address: Ahlatlıbel Mahallesi, 1859. Cadde, No 44/32, Çankaya, Ankara, Türkiye.

You can reach us about anything in this policy (including access, correction and deletion requests) at hello@hellolila.app.

Our representative in the United Kingdom under Article 27 of the UK GDPR is Yasar Yilmazer, 25 Eliot Close, Bristol, BS7 0PZ, United Kingdom. People in the United Kingdom and the Information Commissioner’s Office can contact the representative at uk-representative@hellolila.app. The representative is a contact point only: they do not operate Lila and are not the controller of your data.

Our representative in the European Union under Article 27 of the GDPR is Peren Vardarlıer, Carrer de Calàbria 125, Ppal 4, 08015 Barcelona, Spain. People in the European Union and the EU data protection authorities can write to the representative at that postal address or by email at eu-representative@hellolila.app, a mailbox we operate: we read it and pass to the representative anything that needs them. The representative is established in Spain, where the data protection authority is the Agencia Española de Protección de Datos (AEPD). We name it for reference only, and you can still complain to the authority in your own country. The representative is a contact point only: they do not operate Lila and are not the controller of your data.

2. Data we collect

Almost everything below is data you give us by using Lila. We do not buy data about you, and we do not build a profile of you from other sites.

Some of these only exist if you turn on the feature that needs them. Nothing in this list is collected in the background without a switch you can see.

• Account information. Your email address and name, and the identifier from Google or Apple if you sign in that way. Passwords are stored only as a hash.
• Voice recordings you create, for as long as it takes to transcribe them.
• Reminder content and schedules.
• Chat conversations with Lila, including the replies Lila writes on your phone. When you are signed in, conversations answered on your phone are copied to Lila’s servers (section 11), so your chat history is not split between your phone and our servers. This can include conversations you had on that phone before you signed in. Conversations you mark “Off the record” are never copied.
• Memories and context Lila derives from your conversations. The things it remembers so it can be useful later.
• Workspace content you create: notes, intentions, goals, habits and daily plans.
• Beliefs and practices, only if you tick the separate consent box when you choose them (see section 3). Without that tick nothing about your religion is stored.
• Birth date, and birth time and place if you enter them for astrology features. Time and place of birth are precise enough to identify a person, so we treat them with the same care as section 3 data.
• Photos, files and documents you add yourself, and what Lila reads out of them.
• A photo you pick to give Lila a new look is processed on your phone. Only an outline shape is kept and synced; the photo itself is never uploaded or stored.
• Content from accounts you connect (section 8), when Lila reads it to answer you.
• Places you ask Lila to remember. If you tell it to save your home or your office, that place’s coordinates are stored so the app can recognise when you arrive. Where you are at any moment stays on the phone — the comparison happens there, not here. The exception is a lookup you ask for in a mini-app: for places near you, the phone rounds your position to about 1 km and sends that area to Lila’s servers, which look it up in Lila’s own map data (from Overture Maps) and send it to no other company; for the weather where you are, the rounded area goes through Lila’s servers to MET Norway. Lila’s servers use the area for that one answer and do not store it.
• Device information needed for push notifications, including your notification token, plus a record of which notifications were delivered.
• Subscription status. Payments are taken by Apple or Google; we never see your card details.
• Diagnostics: crash reports and error traces, so we can fix what breaks.
• Usage analytics, which are optional and can be turned off in the app.
• Anything a mini-app you build collects. See section 7.

3. Sensitive data, and the consent it needs

Some information carries extra protection in law: beliefs, health, ethnicity, political opinion, trade union membership, biometrics, sex life and sexual orientation, and — under Turkish law — criminal convictions, appearance and dress, and association membership.

Lila asks for some of this on purpose, because an assistant that does not know you fast during Ramadan, or that schedules a reminder through your prayer time, is not respecting you. But that is our reason, not your obligation.

So the rule is: we do not store any of it unless you have separately and explicitly agreed to that specific category. For religion this is a box you tick yourself, on its own screen, before any choice about faith appears — it is never ticked for you, and it is not part of accepting these terms. We record that you agreed, which version of the wording you saw, and in which language. You can withdraw at any time in Settings → Privacy & Data → Religion data. Withdrawing takes effect immediately: we stop using it, and we delete your faith, your practice settings and your religious reminders.

Health information you record in an app you build inside Lila (for example mood or exercise logs) follows the same rule: a separate consent for that category and that purpose, which you can withdraw in Settings → Privacy & Data.

Two honest caveats. First, if you mention something sensitive in an ordinary chat message, it is stored as part of that conversation, because we cannot detect it reliably enough to strip it out and pretending otherwise would be a false promise. Second, sensitive categories are never used to target advertising, are never sold, and are never shared with anyone outside the providers in section 12.

Every read of data in these categories is written to an internal access log — which category and which record, never the content — so that an audit can show what was looked at and when.

4. How we use your data

Each purpose below is a thing the product actually does. We do not use your content for advertising, and we do not sell it.

• To provide the reminder service and deliver notifications at the right time.
• To transcribe voice recordings and extract what you asked for.
• To assemble your briefings and give Lila the context that makes its answers useful.
• To respect what you have told us about your beliefs and practices, where you have consented to that under section 3.
• To run your account and your subscription.
• To keep the service secure. Rate limiting and abuse prevention.
• To diagnose crashes and errors.
• To answer you when you contact support.
• To see what tends to help people generally, Lila counts whether a kind of Moment was acted on, dismissed, or let pass — grouped by time of day and language, and only where at least fifty people are in the same group. Message text is not used, and the counts hold nothing that identifies you. You can stop contributing in Settings → Privacy & Data; future counts then do not read you, and counts already published hold no identifier of you, so there is nothing in them to remove.

5. AI processing

Lila does not use your content to train its own AI models and does not sell it.

We send your text and audio to AI providers to get an answer back. Today those are Groq, which runs the language model behind chat and transcribes your voice; OpenAI, the fallback for both and the source of the numerical representations that let Lila search your own memories; and Google Gemini, which handles some requests, such as reading the files and photos you add, and also runs the reply-writing model behind a live voice conversation. Each of them processes what we send only to return a result to us, and none of them trains on it. Cloudflare Workers AI is a backup for one step, used only when we switch that backup on: if the usual provider is unavailable, it turns your message into numbers so Lila can choose the right tool. Cloudflare does not train on it. Anthropic prepares the context behind chat replies and, on Pro, designs the apps you ask Lila to build, so it receives the text of your messages and of your app-building requests. It processes that only to return a result to us, under Anthropic’s commercial terms (https://www.anthropic.com/legal/commercial-terms), which do not allow it to train on it, and it keeps it only as those terms allow. Its privacy policy: https://www.anthropic.com/legal/privacy.

Live voice conversation. If you start a live voice conversation (Pro), what you say is sent to OpenAI, which turns your speech into text in real time and, in most calls, turns Lila’s reply back into speech. Google Gemini writes the reply; if it is unavailable, the providers that answer you in chat take over. The live audio and the live captions of the call (the text of what you say and of Lila’s reply) travel in real time between your phone and Lila’s voice service through LiveKit. LiveKit is a real-time transport provider, not an AI provider, and relays them as our processor under its own terms. OpenAI and Google Gemini process what we send only to answer you, and neither trains on it.

Building apps. When you ask Lila to build an app, OpenAI writes its code; on Pro, Anthropic designs it first. Both receive your app request, its specification and its design instructions, which can include personal information you put in them. Each processes them only to return a result to us, and its terms do not allow it to train on them.

Readings in the Observatory (Pro). On Pro, tarot and sky readings are written by the same AI providers that answer you in chat. A card reading carries the cards drawn, the spread and the question you typed, if you typed one. We do not store the question; we keep the text of the reading so it can be shown to you again, and delete it with your account. A sky reading carries today’s moon phase and aspects and your Sun, Moon and rising signs, but never your name, birth date, birth time or birth place. If you give the separate permission on that screen, it also carries a summary derived from your birth details: the sign and house of each planet and the aspects between them (for a Vedic reading: the lagna, the signs, the lunar mansions and the month your running dasha period ends). A partner’s birth details never leave your phone. Without Pro, readings are written on your phone and nothing is sent. You can withdraw the chart permission on the same screen; turning off cloud AI, as described above, stops all of it.

Lila asks before it sends anything of yours to these providers, feature by feature, and you can change your answer at any time in Settings → Privacy & Data → Cloud AI. Section 5A lists each provider, what it receives and what its terms allow.

If you add your own API key for an AI provider (“bring your own key”), requests that use it go to that provider under your account and are billed to you. That provider’s own terms and privacy policy apply to those requests, including its age and region rules. We store your key encrypted and never show it back in full; you can remove it at any time.

You are always told when you are talking with an AI: the chat and every voice screen say so. Share cards and meditation audio that an AI made carry a “Made with AI” label, and the files themselves carry a machine-readable mark that says the same.

AI output can be wrong. Lila is a tool for organising your life, not a source of professional, medical, legal or religious advice.

5A. Third-party AI services

These are all the companies outside Lila that run AI for a feature you use, what each one receives, and what its terms let it do with it. The list comes from the same provider register as the consent screen in the app, and a test fails if the two disagree.

Groq, OpenAI, Google (Gemini) and Anthropic give your data the same or equal protection as this policy. Their API terms let them use what Lila sends only to return a result to us, they do not train their models on it, and they keep it only for as long as those terms allow.

You decide for each feature. When you first open Lila, it shows you each feature that uses cloud AI and sends nothing to these companies until you allow it. You can turn each feature on or off at any time in Settings → Privacy & Data → Cloud AI. Turning a feature off stops new requests to its providers at once; it does not recall what was already sent.

Deleting your account (section 14) deletes everything Lila holds. What a provider has already received is deleted on that provider’s own schedule, shown in the table below.

6. Voice and emotion

If you turn on the emotion feature, Lila estimates a mood, an energy level and a speaking rate from the sound of your voice — not only from your words — so it can adjust its tone.

We are telling you this plainly because it is the kind of thing you should be told plainly. Inferring emotion from a voice is inference from biometric data, and under the EU AI Act a system that does it has to say so to the person using it. That obligation applied from 2 August 2026. Under the EU AI Act this is an emotion recognition system; it is lawful for personal use, and we tell you about it here and in the app before you turn it on. It never creates a voiceprint and is never used to identify you.

What this is not: it is not a diagnosis, it is not a measure of your mental health, it is never used to decide what you are shown or charged, and it is never shared with anyone. The estimate is approximate and is often simply wrong.

You can turn it off, and off means the audio is transcribed and nothing about your tone of voice is derived from it.

7. Apps you build inside Lila

Lila lets you build small apps of your own inside it. That is a genuinely open-ended capability, and we would rather explain how we keep it honest than publish a list of categories that your first app makes out of date.

Before a mini-app can run, it declares which kinds of data it needs, drawn from the same fixed registry of categories this policy uses. You see that declaration and you approve it. An app cannot read a category it did not declare.

If a mini-app asks for a category from section 3, it needs the separate explicit consent described there — building it yourself does not bypass that gate.

Mini-apps you build run for you. We do not use their contents for anything else, and nothing about them is shared with other users unless you deliberately publish it.

8. Accounts you connect

Some data reaches Lila only because you deliberately connect a source. Nothing in this section happens unless you complete that connection, and you can disconnect at any time.

• Google Calendar. If you connect it, we cache your upcoming events so briefings can prepare you for them and reminders can avoid clashing with them. The cache is kept for 30 days. Lila’s use and transfer to any other app of information received from Google APIs will adhere to the Google API Services User Data Policy (https://developers.google.com/terms/api-services-user-data-policy), including the Limited Use requirements. Google data is never used for advertising and never used to train AI models, ours or anyone else’s.
• Notion. If you connect it, Lila reads what you ask it to, and only reads: it does not add, change or delete anything in your Notion workspace. What Lila reads is sent to the AI provider that answers you and is kept with that conversation for as long as the conversation is kept.
• YouTube. When you ask Lila about a YouTube video or channel, it uses YouTube API Services. By using that feature you agree to the YouTube Terms of Service (https://www.youtube.com/t/terms), and Google’s Privacy Policy applies (https://policies.google.com/privacy). Data from YouTube is kept for no more than 30 days. You can revoke Lila’s access to your Google account at any time at https://myaccount.google.com/permissions.
• Spotify. If you connect Spotify to pick an alarm or a playlist, Lila reads your playlists to show them to you. Spotify content is never sent to an AI model, and Lila does not control Spotify by voice; asking Lila by voice opens the Spotify app instead. When you disconnect, Lila deletes the Spotify data it held within 5 days.
• Sources you add yourself, such as a web page to monitor. We fetch and store their content, not anything about you.
• Every access to your Google data is logged for 90 days so you and we can audit it.
• You can disconnect any account in Settings, and deleting your account disconnects all of them.

9. What Lila does not do today

It matters as much what is switched off as what is on. As of the effective date of this policy:

• Lila does not read your calendar to work out whether you are busy and should not be interrupted. That capability exists in the code and is deliberately disabled.
• Lila does not listen for a wake word and does not listen in the background. Recording starts when you start it.
• Lila sends a proactive notification only when it has something useful to say, and only within the limits you set in Settings (how often, and when to stay quiet). You can turn them off there.
• Lila does not track your location continuously. Nothing runs in the background reporting where you are, and your position is never streamed to us.
• Lila does not act on your Gmail on your behalf.
• Lila does not use your content to train its own AI models and does not sell it. Cloudflare Workers AI, a backup used only when we switch it on, does not train on it either (section 5).
• We do not sell your data, and we do not use it for advertising.
• Lila does not create a voiceprint or a face template of you, and a photo you give Lila for a new look never leaves your phone.

10. How long we keep it

Data with a fixed lifetime is deleted automatically when it expires. You do not have to ask.

11. Where your data is stored

Lila’s servers, database and cache all run on Railway in the United States, in the region Railway calls us-west2. If you are in Türkiye, the EU or the UK, your data leaves your country when you use Lila. We would rather say that in one sentence than bury it.

Most of the providers in section 12 are also in the United States. PostHog, which handles the optional usage analytics, is on its European infrastructure. Apple and Google operate globally.

We are working through the formal transfer paperwork each jurisdiction requires — the standard contracts Turkish law asks for, and the clauses EU and UK law ask for. Where that work is not finished we say so here rather than implying otherwise; this section is updated as each one is completed, and the version stamp at the top of this page changes with it.

12. Who processes data with us

We do not sell your data. We share it only with the providers that run the service, when the law requires it, or with your consent. These are the providers Lila uses today.

This table is generated from the service’s own configuration, and a test fails if the two disagree — so it cannot quietly go out of date the way such tables usually do.

When you share something from Lila to Instagram, Facebook, TikTok, Snapchat or another app, the card you chose goes to that app, which handles it under its own terms and privacy policy. Lila sends nothing else, and those companies are not our processors.

13. Your rights

You have the right to access your data, correct it, export it, delete it, and object to processing. Turkish law adds the rights set out in KVKK (Law No. 6698); EU and UK law add the rights set out in the GDPR. We apply all of them to everyone.

Most of this you can do yourself, immediately, without asking us:

• Access and export. Settings → Privacy & Data → Download my data. You get a file containing your memories, reminders, preferences and conversations.
• Correct or delete a memory. Settings → Memory. Open a memory to edit it, or choose to forget it.
• Withdraw consent for a sensitive category. Settings → Privacy & Data. Withdrawal takes effect immediately and deletes what was held under it.
• Withdraw your religion consent. Settings → Privacy & Data → Religion data. This deletes the data at once.
• Opt out of analytics. From the Settings screen, or ask us to do it.
• Delete everything. See section 14.
• If you would rather we did it, email hello@hellolila.app from the address on your account.
• If you think we have got this wrong, you can complain to your data protection authority: the Kişisel Verileri Koruma Kurumu in Türkiye, your national authority in the EU, or the ICO in the UK.

13A. If a breach affects your data

A personal data breach is when data about you is lost, or is seen, changed or destroyed by someone who should not have access to it.

If a breach affects your personal data, we report it to the competent authority within 72 hours of becoming aware of it, where the law requires us to: the Personal Data Protection Board (KVKK) in Türkiye, the competent data protection authority in the EU, or the ICO in the UK.

If you are affected, we also tell you without undue delay, by email to the address on your account. If we cannot reach you by email, we publish a notice on hellolila.app. The message says what happened, which of your data was involved, what we have done about it, and what you can do to protect yourself.

14. Deleting your account

You can permanently delete your Lila account and all associated data in one of two ways:

• In the app: You → Settings → Account → Delete Account. This performs a full deletion of your account and all linked data.
• No access to the app? Email hello@hellolila.app from the address registered to your account. We will delete your account and data within 30 days.
• What gets deleted: your account, reminders, chat conversations, memory and context data, workspace content, mini-apps you built, sensitive data held under section 3, cached calendar and email, notification history, and the link to your subscription.
• Cancelling the subscription itself is done separately through Apple or Google.

Delete your account

15. Age

You must be at least 18 years old to use Lila. It is not intended for children, and we do not knowingly collect data from anyone under 18.

The floor is 18 everywhere, with no country-by-country variation. That is deliberately stricter than the law requires in most places. Lila asks about beliefs, birth data and tone of voice, and it lets you build things that collect more — none of which we want to be defending a thirteen-year-old’s consent to.

If you believe someone under 18 has given us personal data, email hello@hellolila.app and we will delete it.

16. This website

This website sets no cookies and stores nothing on your device. There is no cross-site tracking, no advertising network and no fingerprinting.

We do count anonymous events (a page being viewed, a button being pressed) so we can tell whether what we wrote is understood. Those events carry the page, the language you are reading in, and where the link came from. They do not carry your name, your email address, an identifier that survives the page, or anything you type.

If you join the early-access list, we keep the address you gave us and the language you were reading in, so that we write to you in it. If you tell us in your own words what you would use Lila for, or give us a name to call you by, we keep those as well — both are optional and the list works without them. Where a link tells us which campaign or which site brought you here, we keep that too. If you later create an account with the same address, what you told us here is carried into it, so that Lila can pick the thread up instead of asking you again; you can delete it from inside the app. Every message we send carries a link that removes you, and that link deletes the record and everything you wrote with it, rather than marking it inactive.

If you ask Lila something in the chat on this site, that question — and the last few messages of the same conversation — are sent to a language model running on Cloudflare’s network, so that it can answer in the language you are reading. The model is given passages from this website and nothing about you. We write none of it down: no log, no database entry, no analytics event. The exchange lives in the page and is gone when you close it.

17. Changes to this policy

When this policy changes, the version number and effective date at the top of this page change with it. Material changes are announced in the app before they take effect.

If a change widens what we collect or what we do with it, we ask again rather than assuming your previous agreement carries over.

18. Contact

For privacy questions or data requests: hello@hellolila.app

For anything else: hello@hellolila.app