MasterG Privacy Notice
Version beta-1, 10 November 2026
This notice says what MasterG records about you and your child, why, who sees it, how long it is kept and how to have it deleted. Please read it before you agree. If you would like it explained in Hindi, ask us.
1. Who runs MasterG
1.1 MasterG is run by Akash Parashar, an individual, of Rajapuri, New Delhi 110059, India. There is no company yet.
1.2 "We" means Akash Parashar and the small MasterG team that works with him.
1.3 For questions and complaints, write to team@masterg.app.
2. What the beta is for
2.1 MasterG is in a private beta, by invitation only. The purpose of the beta is to test MasterG with real children and fix what goes wrong.
2.2 So recording the sessions, review by the team, the automated review and keeping samples are all part of the beta. They are not optional. A family that does not accept them cannot join the beta.
2.3 We do not sell personal data or show ads.
3. What we collect
3.1 About your child:
- a. First name and class. The class code, if the child joins a school class. A roll number, if the child types one when joining a class.
- b. Handwriting: a picture of each step the child writes, and the pen strokes behind it.
- c. The text read from each picture, whether the step was judged right, wrong or unsure, and the type of mistake.
- d. What your child says to Masterji, as text.
- e. What Masterji says, as text.
- f. The events of the session and their times, such as when a step was written and when Masterji spoke.
- g. The device: its type, its screen size, and whether the child used a pen, a finger or a mouse.
- h. Masterji's memory of your child (clause 6).
3.2 Sound. During a session, sound from the microphone is sent live to Google so that Masterji can listen and speak. The microphone can pick up other people nearby. We do not keep audio recordings.
3.3 About you, the parent or guardian:
- your email address;
- your name, if you sign in with Google;
- your consent record: who agreed, for which child, each tick, the date and time, and the version of the texts;
- messages you send us.
3.4 About teachers: name, role, email address, school, class name and class level, and the record of the teacher's confirmation.
3.5 Server logs. Our server notes the internet address (IP address), the time and the browser type of each request.
4. Why we use it
4.1 We use it:
- a. to tutor your child: read each step, check it, speak, and pick up where the child left off;
- b. to show progress to you, and to the teacher if the child is in a school class;
- c. to find and fix problems in MasterG (clause 5);
- d. to improve Masterji with samples (clause 7);
- e. to count results in totals that identify no one, such as how many Class 5 children made one kind of mistake.
4.2 We do not use it to advertise or to sell anything to your child.
5. Who looks at sessions
5.1 The team. The MasterG team opens sessions in a private review page. They see the handwriting, the transcript and your child's first name. The team uses AI coding assistants (from Anthropic and OpenAI) to help look into a problem. An assistant then reads the session concerned, including the handwriting pictures and the transcript.
5.2 The automated review. The team runs an automated review, usually each night. Transcripts of sessions are sent to an OpenAI model through the team's own OpenAI subscription. It looks for moments where Masterji went wrong. Before a transcript is sent, software replaces your child's name with a marker. Handwriting pictures are not sent. The name removal is automatic and can miss a name. Other things your child says stay in the transcript.
5.3 The issue tracker. When we find a problem, we record it in the team's private issue tracker, Linear. An issue can hold handwriting crops and short transcript excerpts from your child's session, with the session number, the date and the device type. We need this to see how children write on different devices. Only the team can see it. We do not attach your child's name where we can avoid it. A name can still appear if it is written in the picture, or if the automatic removal misses it.
5.4 Laptop copies. For the nightly review, copies of sessions are kept on team laptops for a short time (table in clause 9).
5.5 The teacher. If your child is in a school class, the teacher sees the progress report of each child in that class.
6. Masterji's memory
6.1 Masterji keeps a short memory for each child: the child's first name, the last topic, and one thing to practise. It holds learning facts only: topics, types of mistake and level.
6.2 You can see the memory in Settings. You can switch it off.
6.3 When the memory is off, Masterji starts each session fresh, and what the memory held is deleted.
7. Samples kept to improve Masterji
7.1 Before a recorded session is deleted, we may copy parts of it as samples: handwriting pictures and transcript text, with what was read and the type of mistake.
7.2 We remove the child's name and the link to the account, the class and the session. A team member checks each sample and drops it if it shows a name or another personal detail.
7.3 After that, we cannot tell which child a sample came from. So we cannot find and remove your child's samples later, even if you delete the account. A sample is still your child's handwriting or words, so a person who knows your child well might recognise it. We keep nothing that says whose it is.
7.4 You can stop new samples. Tell us at team@masterg.app, and we will exclude your child's sessions from samples from then on. Samples already made stay. This request does not stop the team review in clause 5, including crops in the issue tracker.
7.5 Samples are used only to test and improve Masterji. We do not publish them. They are deleted after the period in clause 9.
8. Companies that handle the data for us
8.1 We use these companies:
| Company | What it does for us | What it receives | Where |
|---|---|---|---|
| Google (Gemini API) | Reads the handwriting. Runs Masterji's listening and voice. Writes progress summaries. | Handwriting pictures, live microphone sound, the text of the session, the child's first name | Google's servers, which may be outside India |
| Supabase | Database and sign-in | Accounts, child profiles, classes, progress, consent records | Mumbai, India |
| Amazon Web Services | The server | Recorded sessions, server logs, backups | Mumbai, India |
| Resend | Sends sign-in emails | Your email address | Outside India |
| OpenAI | The nightly automated review | Transcripts with the child's name removed | Outside India |
| Anthropic and OpenAI (AI coding assistants) | Help the team look into a problem | The session concerned: handwriting pictures, transcript, events | Outside India |
| Linear | The team's issue tracker | Handwriting crops, short transcript excerpts, session number, date, device type | Outside India |
8.2 These companies handle the data under their own terms. Where a company offers a setting that stops it using the data to train its own models, we switch that setting on.
8.3 We do not give the data to anyone else, unless the law or a court requires it.
9. How long we keep it
| What | How long we keep it |
|---|---|
| Recorded sessions: step pictures, pen strokes, text read, right or wrong, what was said, events | 90 days from the session, then deleted |
| A recorded session named in an open bug report | Until the bug is closed, 12 months at most |
| Copies of sessions on team laptops | 14 days, then deleted |
| Handwriting crops and transcript excerpts in the issue tracker | Up to 3 years |
| Samples with the name and account link removed | 3 years from the month of the session, then deleted |
| Child profile, progress report and Masterji's memory | While the profile exists. If nobody signs in for 12 months, we email a warning and then delete them |
| Your account (email address, name) | While the account exists |
| Consent record | While the account exists, and 3 years after |
| Server logs | Up to 1 year |
| Backups | Overwritten within 90 days |
| Totals that identify no one | No time limit |
| After a deletion request: data in our live systems | Deleted within 30 days of the request |
| After a deletion request: backups and copies | Deleted within 90 days of the request |
10. Your choices and deletion
10.1 In Settings you can:
- see your child profiles and correct them;
- see Masterji's memory and switch it off;
- see what you agreed to and when;
- delete a child's profile;
- delete your account.
10.2 By writing to team@masterg.app you can also:
- ask for a copy of your child's data, which we send within 30 days;
- ask us to exclude your child's sessions from samples (clause 7.4);
- ask for anything in clause 10.1.
10.3 Withdrawing consent. You can withdraw your consent at any time. Review and samples are part of the beta, so withdrawing means your child stops using MasterG. We then treat it as a deletion request.
10.4 What we delete from our live systems, within 30 days: the child's profile, progress report, Masterji's memory and recorded sessions.
10.5 What we delete from backups and copies, within 90 days: backups, copies on team laptops, and the crops and excerpts in the issue tracker. We find those by searching the tracker for your child's session numbers and first name. We email you when the deletion is done.
10.6 What we do not delete:
- samples already made, because we cannot tell which are your child's (clause 7.3);
- your consent record, kept for the period in clause 9 to show what was agreed;
- server logs, kept for the period in clause 9;
- totals that identify no one;
- what Google and OpenAI hold for a limited time under their own terms. We cannot delete that ourselves.
11. Schools
11.1 A teacher can create a class. Children join with the class code and their first name. They do not need an email address or a password.
11.2 Before a child joins, the school gives the parent the MasterG letter and collects the parent's signed reply. The school confirms this to us. We do not collect the signed replies ourselves unless we ask the school for them.
11.3 A parent whose child uses MasterG through a school has the same choices as in clause 10. Tell the class teacher or write to team@masterg.app.
12. Security
12.1 The connection between your device and MasterG is encrypted.
12.2 Only the team can open the review page and the issue tracker.
12.3 If we learn of a leak that affects you or your child, we will tell you without delay.
13. Questions and complaints
13.1 Grievance contact: Akash Parashar, team@masterg.app, Rajapuri, New Delhi 110059, India.
13.2 We reply within 48 hours and aim to settle the matter within 30 days.
13.3 You also keep any right the law gives you to complain to a consumer forum or to an authority set up by law.
14. Changes to this notice
14.1 Each version has a name and a date. This one is beta-1, 10 November 2026.
14.2 If we change what we collect, why we use it or who receives it, we will ask you to agree again before your child's next session.
14.3 For other changes we will show you the new version in the app.