TentaGen

TentaGen.se

Built by a clinician, for real exams

TentaGen is developed by a practising physician and medical educator, not a general-purpose AI startup. It is designed around how exam questions are actually written, reviewed and imported into university examination systems.

Currently in pilot phase — free to try.

Who builds it

Parviz Mammadzada, MD, PhD, FEBO

Ophthalmologist and eye surgeon, researcher and teacher of medical students at Örebro University, Sweden.

TentaGen came out of his own teaching practice. Writing one good exam question — with a clinical vignette, plausible distractors and an honest cognitive level — takes 20–30 minutes. Doctors are expected to produce assessment, but item-writing methodology is not part of medical training.

The AI drafts, the educator decides.

Every question is reviewed, edited and approved by a human before it reaches a student.

Standards and interoperability

Most AI question generators stop at the text. TentaGen is built to travel through the standards layer into the systems universities actually run their exams in.

Exports

  • IMS QTI 2.1
  • IMS QTI 2.2 (Inspera-compatible)
  • WISEflow JSON (current and legacy)
  • Word / CSV

Pedagogical criteria encoded in the tool

  • Five taxonomy frameworks (Bloom's, NBME, SOLO, Feisel-Schmitz and others)
  • Distractor homogeneity
  • The cover-the-options principle
  • Explicit cognitive-level labelling with a stated rationale

Standards work: TentaGen is exploring 1EdTech's QTI certification pathway. Every export is validated against the published schemas, but a valid package says nothing about what survives the import. So seven representative items were tested in a live WISEflow tenant at a Swedish university: QTI 2.2 took four of the seven item types, QTI 2.1 took all seven but two arrived without an answer key, and the platform's own format took all seven with the scoring intact.

Presented and accepted at academic conferences

TentaGen's approach has been submitted to, and accepted by, academic conferences following review by their programme committees.

Presented

EDEN 2026 Annual Conference — Innovation Demo Lab

Porto, June 2026

Accepted submission, live hands-on demo session.

Upcoming

1EdTech Learning Impact Europe

Thessaloniki, 17 September 2026

”Standards-Based AI for Assessment: Lessons from HE Pilots”

Karolinska Institutet Educational Congress

7–8 October 2026

Accepted hands-on workshop (Scholarship track)

Medicinteknikdagarna (MTD)

Örebro, 13–15 October 2026

Accepted presentation

In use

TentaGen is in use in a live course at a Swedish university. Educators from several countries have also tried it after conference demonstrations, and their feedback has shaped it.

Honest framing, kept deliberately:

  • These are early pilots, not finished validation.
  • Question quality is not perfect out of the box — that is exactly why the review step exists.
  • What is being learned in the pilots feeds directly back into the tool.

Work with TentaGen

Are you an educator?

Try it in your own subject. If it is useful, I will set your department up with a free pilot.

Do you build assessment or LMS platforms?

TentaGen exports to open standards on purpose. If you build an assessment platform, give me access to your importer and I will run the same seven items through it and send you exactly what survived.

Are you researching AI in assessment?

Collaboration on validation of AI-assisted item writing is welcome.

parviz@tentagen.se
🎓 Pilot version — in use with university educators · new pilot partners welcome

💬 Your feedback shapes this tool — tell us what works and what can be improved!

Data privacy: Your questions are stored in your personal database. Uploaded documents and URLs are processed by AI to generate questions and are not stored after generation.

Developed by Parviz Mammadzada, MD, PhD

support@tentagen.se

Distribution for broader use requires permission from the developer.