For administrators & IT
Everything you need to evaluate Authentext in one page — data flow, subprocessors, security controls, and the documents your review will ask for.
One-paragraph summary
Authentext is a teacher tool for writing-authenticity triage. It analyzes only submitted student writing — imported from Google Classroom or uploaded by the teacher — against that student's own writing history. It installs nothing on student devices, monitors no behavior, records no keystrokes, and makes no automated decisions. Its output is evidence for teacher judgment, tier-coded by how far a submission departs from the student's own baseline.
Data flow
Step 1
Google Classroom
Submissions import via OAuth; teacher-scoped; read-only.
Step 2
Authentext (Cloud)
Encrypted storage (AES-256 at rest, TLS 1.2+ in transit); row-level security on every table.
Step 3
Analysis
Deterministic stylometric engine (primary) + Google Gemini API cross-check (enterprise API terms: inputs are not used to train models).
Step 4
Teacher dashboard
Results visible only to the submitting teacher; admin views are pseudonymized.
What we collect / what we never collect
Collected
Student name and class roster (teacher-scoped), submitted writing, derived stylometric metrics, teacher notes and flag resolutions.
Never collected
Keystrokes, screen activity, browsing, location, biometrics, device identifiers, anything from student devices, anything outside submitted schoolwork.
Subprocessors
| Provider | Purpose | Data touched |
|---|---|---|
| Supabase | Database & authentication | Encrypted student writing & metrics |
| Google Gemini API (enterprise terms) | AI cross-check | Submission text — contractually excluded from model training |
| Google (Classroom API) | Assignment import | Rosters & submissions, read-only |
| Lovable (domain DNS via Cloudflare) | Application hosting | Application traffic |
| Resend | Teacher notifications | Teacher email only — no student content |
Security controls (live-verified)
Row-level security on all student tables • forensic tables write-protected from client access • immutable audit logs • pseudonymized admin views with a justification-logged reveal gate • Google tokens never exposed to the client • encryption in transit and at rest.
These controls are continuously verified, not annually asserted — see our live Trust page.
Documents
Privacy Policy • Data Processing Addendum • Terms of Service • What Authentext can't do • Accessibility
Breach commitment: affected schools notified within 72 hours of a confirmed breach involving their data.
Successor clause: our data commitments — no advertising, no sale of student data, no AI training on student work — are written into our Terms to bind any acquirer or successor of Authentext.
The 30-second version for your inbox (the forwardable email)
Teachers: copy this into an email to your principal or tech director.
Subject: Quick approval question — writing-authenticity tool for my classroom
Hi [Name],
I'd like to use a tool called Authentext with my classes this year. It helps me verify writing authenticity by comparing each student's submissions to their own past writing — it doesn't monitor students, doesn't install anything on devices, and doesn't make automated decisions; it just gives me evidence for my own judgment.
Their page for administrators covers data flow, security, subprocessors, and their DPA in one place: authentext.app/admins — and they commit to no advertising, no sale of student data, and no AI training on student work.
It's free during beta, and I'd be the only user unless we decide otherwise. Is this something I can go ahead with, or is there a review step you'd like me to start?
Thanks,
[Teacher name]