Naama

Privacy Policy

Last updated: 18 May 2026

Naama is built and operated by Ariel Goldberg. This policy explains exactly what data the app handles. The short version: we don't collect personal information, sell your data, or run analytics. Your conversations, streaks, and preferences stay on your device.

1. What we do not collect

We do not collect, store, or transmit:

• Personally identifiable information (name, email, phone, address)
• Account or login credentials — we don't have accounts
• Location data
• Photos, videos, or files
• Usage analytics, screen recordings, or behavioral tracking
• Advertising identifiers (IDFA)
• Contact information

We do not run any analytics SDK and we do not have a server-side database that holds personal records.

2. Data that stays on your device

The following information is stored only on your device using Apple's standard iOS storage (UserDefaults). It is never transmitted to us:

• Your conversation history with the AI tutor (Hebrew text + audio files cached locally in your app sandbox)
• Your streak, daily practice time, and progress stats
• Your preferences (hands-free mode, speech speed, etc.)
• Your daily free-tier counter

You can erase all of this at any time by deleting the app.

3. What we send to AI services (and why)

To make the conversation work, the app sends audio and text to AI services through our proxy at sahbak-api.vercel.app:

• Your voice recording is sent to OpenAI's speech-to-text service (gpt-4o-transcribe). It is used to produce a Hebrew transcript and then discarded. We do not retain audio on our servers.
• The Hebrew text of your turn is sent to OpenAI's language model (gpt-5.4-nano) to generate the tutor's reply. Recent turns are sent with each request so the tutor can stay in context. We do not retain transcripts on our servers.
• The tutor's Hebrew reply is sent to Microsoft Azure Speech for text-to-speech synthesis. The resulting audio is streamed back to your device. We do not retain TTS audio on our servers.

We use these third-party AI services solely to power the conversation. We do not share data with them for training or any other purpose, and we operate under their standard API privacy terms.

4. Microphone & Speech Recognition

The app needs microphone access to record your spoken Hebrew. You will be asked for permission the first time you tap the mic. Audio is captured locally, sent to the AI services described above for transcription, and then discarded.

You can revoke microphone permission at any time in iOS Settings → Naama → Microphone.

5. Notifications

If you opt in to notifications, they are scheduled locally on your device to remind you about your daily streak. We do not send push notifications from a server.

6. Children

Naama is suitable for ages 4+. We do not knowingly collect any information from children — because we do not collect information from anyone.

7. Third-party services we use

• Vercel — hosts our API proxy. Vercel logs basic request metadata (IP, timestamp) for ~24 hours for abuse-prevention; we do not access these logs except to investigate outages.
• OpenAI — speech-to-text (gpt-4o-transcribe) and chat completion (gpt-5.4-nano). Standard API privacy applies; OpenAI does not train on API data by default.
• Microsoft Azure Speech — text-to-speech synthesis. Standard Azure Cognitive Services privacy applies.

We do not include advertising networks, marketing pixels, or any behavioral tracking SDKs.

8. Changes to this policy

If we ever change this policy (for example, if a future version adds an account system or paid tier), we will update the "Last updated" date at the top of this page and describe the change.

9. Contact

If you have any question about this policy or your data:

Ariel Goldberg — iarigoldberg@gmail.com