Data sheet for schools — Teaching Record
Last updated: 3 September 2026 · Domain: sijilapps.com
This page is for your school's IT and data-protection people. Everything in it
is taken from the full privacy policy and from the code itself — not
from marketing copy. Every number here has an automated test that fails if the
code drifts away from it.
Governing text: the Arabic version of this page is the binding one; this is
a faithful translation for convenience.
1. The shortest answer to the biggest question
The platform has no field for identified student data. No student name, no student number, no grade attributed to a named student. This is not an operational promise but the shape of the database: no column exists to receive it.
What the teacher produces — a plan, a worksheet, an assessment, a deck — is their own work, not their students' data.
2. What leaves the server, and where it goes
| What | Where | When |
|---|---|---|
| The text of what the teacher asks to generate (topic, standards, instructions) | Anthropic | On each generate |
| The text of a file the teacher uploaded | Anthropic | Only if they press "Read its content" or "Extract its standards" |
| An image the teacher uploads for marking or as evidence | Anthropic | While it is read, then the copy is deleted immediately |
| The database (sensitive fields encrypted before they arrive) | Neon — Frankfurt, Germany | Storage |
| Running the server | Render — Frankfurt, Germany | Runtime |
| The teacher's email and the message text (recovery codes) | Resend — Ireland | On send |
What never leaves: identified student data (it does not exist), passwords (stored as bcrypt hashes), and images after they are read — deleted on every exit path: success, failure, skip and retry.
3. Is our data used to train models?
No. The platform uses Anthropic's commercial API. Inputs and outputs of that API are not used for model training by default, and retained data is never used for training without the account holder's express permission.
We are not Anthropic's partner and are not sponsored by them — we are a customer buying API calls, exactly as we buy hosting. We say so plainly because claiming otherwise would be both easier and false.
4. Three levels your school chooses
The restriction is in your hands, not ours. A school picks one and it applies to every teacher under the same school code; we record your request with its date and sender:
| Mode | What leaves | What is switched off |
|---|---|---|
| Open (default) | As in §2 | Nothing |
| No images | Text only | Marking from a photo, and photo evidence |
| No content | Topic and standards only | All of the above, plus reading teachers' files and extracting objectives from them |
Under "no content" the platform still works: planning, worksheets, assessments, the coverage record and the classroom tools are all built from the teacher's standards and the topic they type.
To request it: support@sijilapps.com from a school address, with your school code in the platform.
5. What you can see for yourselves
In the school administrator's dashboard, the "What leaves for the model" section: the number of calls over thirty days broken down by task type (plan, worksheet, assessment…), and your school's current mode stated in plain words.
We do not show the content of what was sent — because we do not store it. That is itself part of the answer: what is not stored cannot leak and cannot be demanded in a dispute.
The dashboard is deliberately aggregate: it shows no teacher's name and no individual figures. A dashboard that says "this teacher has three gaps" becomes a monitoring tool, and the first teacher who sees it stops the school's adoption of the tool entirely.
6. Images — the condition we ask of you
Before every image upload the platform states: do not upload personal photographs of students, and cover the names visible on their papers. Complying with your school's policy on images and consent is a precondition, and the responsibility for choosing what to upload is the teacher's.
For anyone who wants a guarantee rather than a reminder: "no images" closes the door at the server, so it no longer depends on a teacher's judgement.
7. Security and retention
- Encryption in the database: phone number, the content of what is generated, evidence descriptions, and uploaded files (split into chunks, each encrypted separately) — with a key the database does not hold.
- Passwords: bcrypt, never stored as text.
- Session: four hours, cookie over HTTPS only. Two-step verification available.
- Sign-in attempt log: a fingerprint, not an address or an email, erased after 30 days.
- Breach notification: within 72 hours of learning of it, as required by Qatar's Personal Data Privacy Protection Law (No. 13 of 2016).
- Deleting a teacher's account erases what belongs to them from twenty tables, not the account row alone.
8. What we do not promise
- We do not promise the data never leaves Qatar. The servers are in the European Union and the model is outside it. If you need nothing at all to reach a model, "no content" is the only honest answer.
- We claim no certification we do not hold — no ISO, no SOC 2 today. The platform is currently one person's work, with a Qatari company being formed (§1 of the policy).
- And we do not say the AI does not err. Everything generated is a first draft the teacher reviews; automatic marking is a suggestion carrying a confidence score, and anything below it is shown as a question, not a result.
9. Contact
support@sijilapps.com — a human reads it, and it is the same address for privacy questions, deletion requests, and requests to change a school's mode.