Privacy Policy

Account information: your name and email address if you sign up with email/password or a social provider (Google or Apple); no email is required for a guest/anonymous account.

Learning data: the words you save, plant, review, and harvest; your spaced-repetition schedule and review history; practice and exercise results; streaks and tasks.

Video learning data: video URLs you add, the captions/transcripts fetched for them, and your watch/shadowing progress.

Browser extension data: when you actively use the extension, the text you select and the target language, sent to our servers to look up or resolve that word.

Voice data: if you use Mầm Talk (spoken conversations with Mầm), the audio from your microphone during a conversation, the transcript of what you and Mầm said, and — only if you allow recording — short recordings of your own turns and the pronunciation scores made from them.

Technical data: standard web request information (such as IP address and browser user agent) collected by our hosting infrastructure for security, abuse prevention, and performance.

Purchase data: if you buy Engrove+ or Xu, the purchase is processed by the relevant app store or payment processor; we receive confirmation of the purchase but do not receive or store your full payment card details.

We use the information above to:

• Provide and operate the Service, including syncing your progress across devices.
• Personalize your spaced-repetition schedule and learning suggestions.
• Power AI-assisted features, such as word lookups from the browser extension and premium learning insights.
• Maintain the security and integrity of the Service and prevent abuse.
• Communicate service updates, and respond to your support requests.
• Understand how the Service is used so we can improve it.

Some features send a limited, specific piece of data — such as a word or phrase you look up, or a summary of your recent learning activity for a premium insight — to a third-party AI provider (currently Anthropic and a ModelStack-hosted model) solely to generate the response you requested. We do not use this data to train our own models, and these providers process it under their own data-handling terms.

Mầm Talk lets you practise speaking English with Mầm out loud. It uses your microphone only while a conversation is open, and only after you allow microphone access on your device. Mầm Talk is intended for learners aged 16 and over.

During a conversation your voice is streamed to our own conversation server, which sends it to Soniox to be turned into text. That text, together with your first name, your learning level, a few of your garden words, your usual mistakes, a short note of something you said in a past conversation, and the day's topic, is sent to Groq, which runs the AI model that writes Mầm's reply. The reply text is then sent to Soniox to be turned into speech. When a direct connection isn't possible, the audio passes through Cloudflare's TURN relay. We keep the transcript of each conversation so you can look back at it. After a conversation, and for messages you type in Mầm's chat mode, the text you said or typed is also sent to a ModelStack-hosted AI model to check your English and write your practice report.

Recording is a separate choice. Before your first conversation we ask whether Mầm may record your voice to give you pronunciation feedback. If you agree, recording is on by default: short clips of your own turns (never Mầm's) are stored privately and scored by our self-hosted pronunciation server. Only you can play them back. If you say no, Mầm Talk still works, but nothing is recorded or scored. Even without recording, your voice is still processed live by Soniox to turn it into text during the conversation; the audio itself is not kept, only the transcript.

Recordings are deleted automatically 30 days after the conversation; the transcript and pronunciation scores are kept with your learning data. You can turn recording off at any time in Mầm Talk's settings, and deleting your account deletes your conversations and any recordings.

We do not sell your personal information. We share information only with the service providers that help us run Engrove, and only as needed to provide the Service:

• Cloudflare — hosting, content delivery, and file storage for the Service.
• Our database provider — stores your account and learning data.
• Google and Apple — only if you choose to sign in with one of these providers.
• Anthropic and ModelStack — process the specific text described in Section 3, for AI-assisted features, and ModelStack also checks the text of your Mầm Talk conversations and typed chats for your practice report (Section 4).
• A video-transcription service — used to fetch captions/transcripts for a video URL you provide.
• Soniox — converts your speech to text during a Mầm Talk conversation, and turns Mầm's replies into speech (for the voice it receives Mầm's text, not your voice).
• Groq — runs the AI model that writes Mầm's replies during a Mầm Talk conversation. It receives the text of what you said and the learner details listed in Section 4, not your audio.
• Cloudflare Realtime TURN — relays a conversation's encrypted audio when your network can't connect to our conversation server directly.
• Our own servers — the Mầm Talk conversation server and the pronunciation-scoring server are run by us, not by a third party.

We may also disclose information if required to by law, or to protect the rights, property, or safety of Engrove, our users, or the public.

We use a session cookie to keep you signed in. We do not currently use third-party advertising trackers or cross-site tracking cookies.

We keep your account and learning data for as long as your account is active, so your progress is there when you come back. If you delete your account, we delete or anonymize your personal data within a reasonable period, except where we are required to retain it for legal or security reasons.

Mầm Talk voice recordings are the exception: they are deleted 30 days after the conversation they come from, as described in Section 4.

You can export a copy of your data from Settings at any time. You can request deletion of your account and associated data by emailing privacy@engrove.app or using the in-app delete-account option, where available. Depending on where you live, you may have additional rights to access, correct, or restrict the use of your personal data — contact us to exercise them.

The Service is not directed to children under 13, and we do not knowingly collect personal information from children under 13. If you believe a child under 13 has provided us with personal information, contact privacy@engrove.app and we will delete it.

Mầm Talk, our spoken-conversation feature, is only for learners aged 16 and over, because it processes voice, and we ask you to confirm that you are 16 or older before your first call. If you believe someone under 16 has used Mầm Talk, contact privacy@engrove.app and we will delete their Mầm Talk conversations and any recordings.

We use reasonable technical and organizational measures to protect your information. No method of transmission or storage is 100% secure, and we cannot guarantee absolute security.

Because we use infrastructure and AI providers that operate internationally, your information may be processed in a country other than the one you live in, which may have different data protection laws.

The Engrove browser extension does nothing on web pages until you read the welcome page that opens after install and choose "Agree and start". You can withdraw at any time by removing the extension.

What it sends to Engrove: the text you select (up to 200 characters), only when you press Translate or the add button, or use the right-click menu; your sign-in session cookie, so your saved words reach your account; and, on a YouTube watch page while YouTube subtitles are on, that video's ID, title and caption text, so we can build bilingual subtitles. If a video has no captions and you choose to have Engrove transcribe it, the video ID is sent to our servers.

What it does not send: bilingual page reading is translated on your device by Chrome's built-in translator; the text of the pages you read, the pages you visit and your browsing history are not sent to us.

We use this data only to provide the extension's single purpose, which is to look up and save vocabulary from the web, and the features you see in it. We do not sell it, we do not use it for advertising or to decide credit, and we do not let people read it except to protect the Service, comply with the law, or with your consent. The extension's use and transfer of user data adheres to the Chrome Web Store User Data Policy, including its Limited Use requirements.

The extension asks for these permissions: storage (your settings), context menus (the right-click commands), alarms (a periodic refresh of the seed count on the toolbar icon), offscreen documents (to play word pronunciations), access to all web pages (to show the selection bubble where you select text), and access to youtube.com (for YouTube subtitles).

We may update this Privacy Policy as the Service evolves. If we make a material change, we will make reasonable efforts to notify you before it takes effect.

Questions about this Privacy Policy or your data? Contact us at privacy@engrove.app.