Privacy Policy

Effective Date: March 26, 2026

Last Updated: November 8, 2026

This Privacy Policy describes how Marejo Life Memory LABS - FZCO ("we," "us," or "our") collects, uses, shares, and protects your personal information when you use the Marejo application ("Marejo" or the "App") on iOS, Android, and the web at app.marejo.io. Marejo is a voice-first personal assistant that combines diary, calendar, reminders, to-do lists, notes, meeting records, document understanding, and AI chat into a single conversation interface.

• November 8, 2026: The operating company changed to Marejo Life Memory LABS - FZCO, registered in Dubai Silicon Oasis (IFZA), United Arab Emirates. This revision updates the company name, the contact details and the statements about where the company is registered. The rights described in this policy did not change.
• October 15, 2026: This revision corrects the short-voice model name to match the code: gpt-4o-transcribe, with one fallback to whisper-1 if that model is unavailable. This is a correction of fact. The November 3, 2026 start date for recipients newly named on October 4, 2026 is unchanged.
• October 4, 2026: This revision corrects the short-voice model name, moves image generation off Google, and names the OpenAI models used as the last chat fallback, for meeting summaries, recurring-request reports, chat titles, and tool-call repair. It also names Azure OpenAI Realtime (live voice), Upstage Document Parse (uploaded documents), Neon (database), Cloudflare R2 (files), and Upstash QStash (a recurring-request run id only). New recipients and new overseas transfers apply from November 3, 2026.
• September 23, 2026: This revision names TypeSafe (Jev) for judgments that choose a saved record, a meeting passage, or whether a memory is worth keeping, and states that a later question about a document sends the extracted text again. It also names TypeSafe (Jev) for meeting-note classification: meeting-transcript excerpts and short claim sentences go through the Vercel AI Gateway so Jev can classify grounding, faithfulness, meeting kind, chapters, and which speaker. Jev does not write the note. That helper does not send a zero-data-retention flag today.
• September 8, 2026: This revision updates the processing items to match the live product. The original effective date is unchanged.

By using Marejo, you agree to the collection and use of your information as described in this policy. If you do not agree, please do not use the App.

For questions or concerns about this policy, contact us at hello@ridiocompany.com.

Data Collection Summary

We do not use product-analytics tracking. PostHog is used for crash and error monitoring only. RevenueCat is not an active data recipient because paid subscriptions are not currently sold.

1. Information We Collect

1.1 Information You Provide Directly

Account Information. When you create an account, we collect your name, email address, and profile image. You may sign in using Google, Apple, or email and password. For Google or Apple sign-in, we receive basic profile information from that provider.

Short Voice Input (Chat and Diary). When you use voice input in chat or diary, your device records audio through the microphone. That audio is transmitted to OpenAI's gpt-4o-transcribe model for transcription into text. If that model is unavailable, the audio is sent once to whisper-1. We store the resulting text transcription. These short audio recordings are not stored on our servers.

Live voice. In live mode, call audio and the instructions for that call go directly from the device to Azure OpenAI Realtime. Speech during the call is transcribed with gpt-4o-mini-transcribe on the same Azure resource.

Meeting Recordings. When you use Meeting Mode, capture is one continuous audio file stored in Cloudflare R2 at a path such as meetings/{userId}/{meetingId}/audio.m4a. That audio is sent to ElevenLabs Scribe v2, which returns a diarized transcript (speakers). OpenAI gpt-5.4, via the Vercel AI Gateway, then writes a title, summary/tldr, decisions, and action items. Meeting audio is user content. It is stored, and it is deleted when you delete the meeting or your account (database cascade plus object storage).

Diary Entries. When you share experiences through voice or text, we save both your original transcription (raw text) and an AI-polished version created by Google Gemini. We also store mood classifications and any photos you attach.

Notes. You may create markdown notes. We store the title, body, and any inline images.

Documents. When you upload a document for understanding, we store the file, file name, extracted text, and an AI-generated summary.

Calendar Events. When you create, edit, or delete calendar events through conversation, we store event details including title, description, start and end times, and reminder preferences.

Reminders. When you ask to be reminded of something, we store the reminder message, scheduled time, and delivery status.

To-Do Items. When you create tasks, we store the title, description, due date, priority level, and completion status.

Photos. When you attach photos to diary entries, notes, or chat, we upload them to Cloudflare R2 and store the file reference and any captions.

User Preferences. You can customize your AI assistant's name (default: "Marejo"), how the AI addresses you (default: "Friend"), chat background color, timezone, locale, notification preferences, quiet hours, custom mood labels, and text-to-speech voice settings.

Memories and Assistant Profiles. We may store memories extracted from your conversations and assistant profile settings (name, how the assistant addresses you, and related persona settings).

Chat History and Usage Metadata. We store your full conversation history with the AI assistant, including messages, AI responses, and tool invocations. For each AI response, we also store technical metadata: the AI model used, provider name, token counts (input, output, reasoning, cached), and raw provider response data. This metadata is used for internal usage metering.

Credit Balance. We may maintain an internal per-user credit balance used to meter AI features. Credits are not currently sold, and all features are available without payment.

Daily Diary Summaries. The App automatically generates AI-powered daily summaries of your diary entries, including an overall mood assessment. These summaries are stored in our database.

AI-Generated Images. When you request image creation (e.g., "draw me a picture of..."), your text prompt is sent to OpenAI gpt-image-2.5-flare via the Vercel AI Gateway. The generated image is stored in Cloudflare R2. Image generation may be subject to a daily limit.

1.2 Information Collected Automatically

Session Data. When you sign in, we record your IP address and user agent (browser/device identifier) for security and session management purposes.

Device and Crash Information. We use PostHog for crash and error monitoring, which may collect device type, operating system version, app version, crash logs, and stack traces when errors occur. We do not use PostHog for product analytics, advertising, or App Tracking Transparency (ATT) tracking. The ATT prompt previously used for analytics no longer exists.

Push Notification Tokens. If you enable push notifications, we store your Expo push notification token to deliver reminders, calendar alerts, and to-do notifications.

1.3 Information from Third-Party Sign-In

When you sign in with Google or Apple, we receive your name, email, and (where provided) profile image from that provider. We also receive and securely store OAuth tokens necessary to maintain your authenticated session.

1.4 Authentication Methods

We support the following authentication methods:

• Google Sign-In — OAuth tokens are stored securely
• Apple Sign-In — OAuth tokens are stored securely
• Email and password — passwords are stored using secure hashing algorithms; they are never stored in plain text

Transactional emails (verification, password reset, account deletion confirmation) are delivered through our email service provider (currently Resend). Your email address and the email content are shared with the email provider for delivery purposes only.

2. How We Use Your Information

We use the information we collect to:

• Provide the App's core features — diary recording, calendar management, reminders, to-do lists, notes, meeting records, document understanding, and AI conversations
• Process short voice input — send chat and diary audio to OpenAI gpt-4o-transcribe for Korean and English speech-to-text transcription, or once to whisper-1 if that model is unavailable (this audio is not stored by us)
• Process meeting recordings — store meeting audio, send it to ElevenLabs Scribe v2 for a diarized transcript, and generate title, summary, decisions, and action items
• Power AI conversations — send your messages, relevant diary entries, notes, meetings, documents, and calendar data to Google Gemini (via Vercel AI Gateway) to generate contextual AI responses. If that model cannot answer, the same gateway sends the turn to OpenAI gpt-5.4
• Generate text-to-speech audio — send text to ElevenLabs to produce spoken audio responses based on your TTS preferences
• Deliver push notifications — send reminders, calendar alerts, and to-do notifications through Expo's push notification service
• Perform web searches — when the AI needs current information to answer your questions, it may use Tavily to search the web on your behalf
• Authenticate your account — verify your identity through Google Sign-In, Apple Sign-In, or email and password
• Deliver transactional emails — send verification codes, account deletion confirmations, and email change confirmations through our email delivery provider (Resend)
• Meter usage internally — track internal usage of AI features; currently all features are available without payment
• Personalize your experience — apply your preferred AI name, nickname, timezone, memories, and other settings
• Monitor for errors — detect and fix crashes and technical issues using PostHog
• Protect security — prevent unauthorized access, detect fraud, and enforce our terms of service
• Enforce terms of service — moderate accounts (including suspension or banning) when necessary to enforce our terms; ban status, reason, and expiration may be stored

Home-screen widgets display existing diary, calendar, and reminder content on your device. They do not collect a new category of personal information.

3. How We Share Your Information

We do not sell your personal information. We share your data only with the following categories of service providers, and only to the extent necessary to provide the App's features:

3.1 AI Processing Services

3.2 Authentication Providers

3.3 Infrastructure Services

3.4 Email Delivery

3.5 Crash and Error Monitoring

We do not share data with RevenueCat while paid subscriptions are not sold.

3.6 Legal Disclosures

We may disclose your information if required by law, legal process, or government request, or to protect our rights, safety, or property, or that of our users or the public.

4. Data Flows

To help you understand how your data moves through the App:

• Short voice input (chat/diary): Your device microphone captures audio, which is sent to OpenAI's gpt-4o-transcribe model, or once to whisper-1 if that model is unavailable. The transcribed text is returned to our server and stored in our database. This short audio is not stored on our servers.
• Live voice: Call audio and the instructions for that call go directly from the device to Azure OpenAI Realtime. Speech during the call is transcribed with gpt-4o-mini-transcribe on the same resource.
• Meeting Mode: Your device captures one continuous audio file. That file is uploaded to Cloudflare R2 and sent to ElevenLabs Scribe v2. The diarized transcript is stored. OpenAI gpt-5.4, via the Vercel AI Gateway, writes a title, summary/tldr, decisions, and action items, and Jev judges the result. Meeting audio is stored as user content until you delete the meeting or your account.
• AI chat: Your text messages are routed through Vercel's AI Gateway to Google's Gemini API along with relevant conversation history and your preferences. If that model cannot answer, the same gateway sends the turn to OpenAI gpt-5.4. The AI response is streamed back to your device. Token usage and model information may be recorded for internal metering. Chat titles use OpenAI gpt-4.1-mini. A failed tool call may be repaired with OpenAI gpt-4o.
• Notes: Markdown content is stored in our database. Inline images are uploaded to Cloudflare R2.
• Documents: Uploaded files are stored in Cloudflare R2. Text extraction may send the file to Upstage Document Parse. Extracted text and an AI summary are stored in our database.
• Photos: Images from your camera or photo library are uploaded to Cloudflare R2. The storage URL is saved in our database and linked to your diary entry or note.
• Push notifications: Your Expo push token is stored on our server. When a reminder or calendar alert is due, our server sends the notification through Expo's push notification service to your device.
• Text-to-speech: When you request spoken responses, text is sent to ElevenLabs, and the generated audio is streamed back to your device.
• Image generation: When you request an image, your text prompt is routed through Vercel's AI Gateway to OpenAI gpt-image-2.5-flare. The generated image is uploaded to Cloudflare R2, and the URL is returned to your device.

5. International Data Transfers

Marejo Life Memory LABS - FZCO is registered in Dubai Silicon Oasis (IFZA), United Arab Emirates. Users in the United Kingdom, the European Economic Area and Switzerland have the rights described in section 8.2. Our services primarily target users in South Korea. We do not have a Korean business registration number or mail-order business (통신판매업) registration number.

Your data may be transferred to and processed in countries other than your own, including the United States, where our third-party service providers operate. The following providers receive data internationally:

For users in the United Kingdom: We process your data in accordance with the UK GDPR and the Data Protection Act 2018. You may complain to your local data protection authority, which in the UK is the Information Commissioner's Office (ICO).

For users in the European Economic Area (EEA) or Switzerland: We rely on Standard Contractual Clauses (SCCs) approved by the European Commission, or other appropriate safeguards, to ensure adequate data protection for international transfers.

For users in South Korea: International transfers are conducted in compliance with the Personal Information Protection Act (PIPA), with appropriate protective measures in place per Article 28-8.

6. Data Retention

We retain your personal information for as long as your account is active and as needed to provide you with the App's services:

Account Deletion. When you delete your account, all associated data — diary entries, notes, meeting records (audio, transcript, and notes), documents, memories and assistant profiles, calendar events, reminders, to-do items, daily summaries, chat history, usage metadata, credit balance, preferences, and session data — is permanently deleted from our database through cascading deletion. Photos, note images, documents, meeting audio, and AI-generated images stored in Cloudflare R2 may take up to 30 days to be fully purged from storage.

Diary Entry Deletion. When you delete an individual diary entry, the text content is deleted immediately. However, photo attachments linked to that entry may persist as standalone files until you delete them separately or delete your account (at which point all photos are deleted).

Meeting Deletion. When you delete an individual meeting, the transcript, summary, decisions, action items, and the stored audio file are deleted (database cascade plus object storage). Object-storage purge may take up to 30 days.

7. Data Security

We implement the following measures to protect your personal information:

• Encryption in transit: All data transmitted between your device and our servers uses HTTPS/TLS encryption.
• Authentication security: We support OAuth 2.0 for Google Sign-In and Apple Sign-In, and secure password hashing for email/password authentication.
• Access controls: Each user can only access their own data. All database queries are scoped to your user ID.
• Input validation: All data inputs are validated using schema validation to prevent malformed data. Database queries use parameterized statements to prevent injection attacks.
• Rate limiting: AI and transcription endpoints are rate-limited to prevent abuse.
• Error monitoring: We use PostHog to detect and respond to security-relevant errors promptly.
• Administrative access controls: Our administrative team may access user accounts for support and security purposes (such as account moderation). Administrative access is logged and limited to authorized personnel. Administrators may, in limited circumstances, access user sessions for support purposes (session impersonation), but this is restricted to customer support and security investigations.

No system is completely secure. While we take reasonable measures to protect your information, we cannot guarantee absolute security. If you discover a security vulnerability, please contact us at hello@ridiocompany.com.

8. Your Rights

8.1 All Users

Regardless of your location, you have the right to:

• Access your personal data stored in the App
• Correct inaccurate information through the App's settings
• Delete your account and all associated data, including meetings (audio, transcript, and notes), notes, and documents
• Manage push notification preferences through the App's settings
• Export your data by contacting us at hello@ridiocompany.com

8.2 United Kingdom, European Economic Area, and Switzerland (UK GDPR / EU GDPR)

If you are located in the UK, EEA, or Switzerland, you have the following rights under the UK GDPR or the EU GDPR, whichever applies to you:

• Legal basis for processing: We process your data based on (a) performance of our contract with you (providing the App's services), (b) your consent (optional features such as notifications or TTS), and (c) our legitimate interests (security, error monitoring, service stability).
• Right to restrict processing: You may request that we limit how we use your data.
• Right to data portability: You may request a machine-readable copy of your data.
• Right to object: You may object to processing based on legitimate interests.
• Right to withdraw consent: You may withdraw consent for optional processing at any time.
• Right to lodge a complaint: You may file a complaint with your local data protection authority.

To exercise these rights, contact us at hello@ridiocompany.com. We will respond within one month as required by law.

You may also lodge a complaint with your local supervisory authority. For UK residents, that is the Information Commissioner's Office (ICO): Wycliffe House, Water Lane, Wilmslow, Cheshire, SK9 5AF. Tel: 0303 123 1113. Website: ico.org.uk.

8.3 California Residents (CCPA/CPRA)

If you are a California resident, you have additional rights under the California Consumer Privacy Act (CCPA) and California Privacy Rights Act (CPRA):

• Right to know: You may request details about the categories and specific pieces of personal information we collect.
• Right to delete: You may request deletion of your personal information.
• Right to opt-out of sale: We do not sell your personal information.
• Right to non-discrimination: We will not discriminate against you for exercising your privacy rights.

To exercise these rights, contact us at hello@ridiocompany.com.

8.4 South Korean Residents (PIPA)

If you are located in South Korea, you have rights under the Personal Information Protection Act (개인정보 보호법):

• Right to access: You may request access to your personal information.
• Right to correction: You may request correction of inaccurate information.
• Right to deletion: You may request deletion or suspension of processing of your personal information.
• Right to portability: You may request your data in a portable format.
• Right to refuse automated decision-making: You may object to decisions made solely by automated processing, including AI-based profiling and meeting summarization.

For detailed information in Korean, please refer to our Korean-language privacy policy (개인정보 처리방침) available at the same URL with the Korean locale.

To exercise these rights, contact us at hello@ridiocompany.com. We will respond within 10 days as required by PIPA.

9. Children's Privacy

Marejo is not intended for children under the age of 14. We do not knowingly collect personal information from children under 14. If you believe we have collected information from a child under 14, please contact us at hello@ridiocompany.com, and we will promptly delete the information.

This age threshold aligns with the Korean Personal Information Protection Act (PIPA), which requires parental consent for processing personal information of children under 14.

10. Third-Party Links and Services

The App may contain links to third-party websites or services. We are not responsible for the privacy practices of those third parties. We encourage you to read their privacy policies before providing them with your information.

11. Push Notifications

Marejo uses Expo's push notification service to deliver reminders, calendar alerts, and to-do notifications. You can manage notification preferences in the App's settings, including:

• Enabling or disabling all notifications
• Enabling or disabling specific categories (reminders, calendar alerts, to-do alerts)
• Setting quiet hours during which notifications are suppressed

Your Expo push notification token is stored on our server solely for the purpose of delivering notifications you have requested.

12. Cookies and Local Storage

The Marejo mobile app does not use cookies. However, the App may use secure local device storage to store session tokens, user preferences, and cached data for performance. This data remains on your device and is not shared with third parties.

If you use Marejo on the web (app.marejo.io) or access legal pages on our website, standard web cookies may be used for session management, subject to your consent where required by law. For details, see our Cookie Policy.

13. Changes to This Policy

We may update this Privacy Policy from time to time. When we make material changes:

• We will update the "Last Updated" date at the top of this policy.
• For standard changes, we will notify you through the App (via push notification or in-app notice) at least 7 days before the changes take effect.
• For material changes that expand the scope of data collection, introduce new third-party sharing, or otherwise significantly affect how we process your data, we will provide at least 30 days' notice and seek your consent where required by applicable law.

Your continued use of the App after the effective date of changes constitutes your acceptance of the updated policy.

14. Supplemental Notices

AI-Generated Content

Marejo uses artificial intelligence to process your conversations, polish diary entries, classify moods, summarize meetings and documents, extract memories, and provide responses. The default chat model is Google Gemini, via the Vercel AI Gateway. If that model cannot answer, the same gateway sends the turn to OpenAI gpt-5.4. Meeting summaries and recurring-request reports use OpenAI gpt-5.4. Chat titles use OpenAI gpt-4.1-mini. Tool-call repair uses OpenAI gpt-4o. Images you ask for use OpenAI gpt-image-2.5-flare. Uploaded documents may be sent to Upstage Document Parse for text extraction. AI-generated content may not always be accurate. Please refer to Google's AI privacy practices for how Google handles data received through its API.

Judgments that choose a saved record, a meeting passage, or whether a memory is worth keeping may be made by TypeSafe Jev through the Vercel AI Gateway. Jev returns a yes/no or a choice. It does not write the message you read. Those requests set zero data retention. Except as needed for legal duties, TypeSafe does not keep prompts longer than it takes to produce the answer. When a meeting is summarized, TypeSafe Jev also receives meeting-transcript excerpts and short claim sentences through the Vercel AI Gateway to classify grounding, faithfulness, meeting kind, chapters, and which speaker. Jev does not write the note. That helper does not send a zero-data-retention flag today.

Voice Data

Voice processing differs by feature:

• Chat and diary: Short audio is sent to OpenAI's gpt-4o-transcribe model, or once to whisper-1 if that model is unavailable. According to OpenAI's API data usage policy, data sent through the API is not used to train their models and is retained for a limited period for abuse monitoring purposes only. We do not store these short audio recordings on our own servers.
• Live voice: Call audio goes directly to Azure OpenAI Realtime. In-call speech is transcribed with gpt-4o-mini-transcribe on the same resource.
• Meetings: Meeting audio is stored in our object storage and sent to ElevenLabs Scribe v2 for a diarized transcript. Meeting audio is user content and is deleted when you delete the meeting or your account.

Biometric Data

Marejo does not collect or process biometric identifiers or biometric information. Voice recordings are used solely for text transcription (and, for meetings, for stored user content and summarization). They are not used for voice identification or voiceprint creation.

Automated Processing

Marejo uses automated processing to create diary entries, calendar events, reminders, to-dos, daily summaries, meeting titles and summaries (including decisions and action items), document summaries, and memories. You can review, edit, or delete these outputs in the App, and you may object as described in Section 8.

Judgments that choose a saved record, a meeting passage, or whether a memory is worth keeping may be made by TypeSafe Jev through the Vercel AI Gateway. Jev returns a yes/no or a choice. It does not write the message you read. Those requests set zero data retention. Except as needed for legal duties, TypeSafe does not keep prompts longer than it takes to produce the answer. When a meeting is summarized, TypeSafe Jev also receives meeting-transcript excerpts and short claim sentences through the Vercel AI Gateway to classify grounding, faithfulness, meeting kind, chapters, and which speaker. Jev does not write the note. That helper does not send a zero-data-retention flag today.

15. Contact Us

If you have questions, concerns, or requests regarding this Privacy Policy or our data practices, please contact us:

• Email: hello@ridiocompany.com
• Phone: +1 (672) 969 2092 / +82 10 9878 2092
• Address: 800-IFZA(4392) Business Park, Dubai Silicon Oasis, Dubai, UAE

For users in South Korea, our Personal Information Protection Officer (개인정보 보호책임자, CPO) is Sang Jin Lee (CEO). Contact details are provided in the Korean-language privacy policy.

---

This Privacy Policy was last updated on October 15, 2026. The October 15 change corrects the short-voice model name to gpt-4o-transcribe, with one fallback to whisper-1. New recipients and new overseas transfers named on October 4, 2026 apply from November 3, 2026. The October 15 correction does not move that date.