How AiQuran works

Ai Quran (AI Quran) helps you study. Scholars and named sources ground what you learn. AI synthesises only from retrieved material. This public trust page is educational support — not a fatwa — and explains what the product will and will not do.

AI will never

  • Invent Quran verses or rewrite Quran text
  • Invent hadith or fabricate citations
  • Present an AI opinion as a fatwa
  • Hide scholarly disagreement
  • Pretend certainty where the retrieve pack is thin

AI will

  • Quote or point to retrieved Quran, named tafsir, and numbered hadith
  • Distinguish Quran wording, tafsir, hadith, and AI synthesis
  • Abstain when evidence is insufficient
  • Direct high-stakes personal questions to a qualified local scholar

Ask a scholar instead

Personal rulings on marriage, divorce, inheritance, aqeedah disputes, and complex fiqh are outside this product. Live scholar booking is not available here.

How this page relates to AI Quran

Searchers looking for AI Quran, Quran AI, or a Quran learning companion should read this policy beside the What is AI Quran page. The reader stays free. Cited AI is a short trial. Methodology stays unlisted because it is a laboratory disclosure, not a marketing page.

Source stack and fatwa boundary

Cited answers show Quran keys, named tafsir, and numbered hadith before labelled AI synthesis. Personal fiqh questions stay in explore-sources mode: the product will not invent a ruling. Recitation remains pacing hints, not Tajweed AI.

How the evidence is built

Evidence first: verses, named tafsir, and hadith sit under the answer before any AI synthesis, with coverage marked Complete or Partial rather than a trust-score percentage. Retrieve-then-cite: the model only writes over material actually fetched for that question. Canonical snapshot, not a live scrape: Uthmani Arabic and Sahih International are stored locally and checksum-verified, with Quran Foundation used to sync and fill gaps. No vector database as source of truth: semantic search returns candidate verse keys only, and the Arabic text and citations shown always come from the canonical snapshot by that key. The Quran Knowledge Graph is typed, named edges between ayahs, not similarity scores from an opaque model. A published evaluation suite checks answers against a golden set on an ongoing basis. The retrieval pipeline itself, the embedding and reranking models, and the evaluation methodology stay in the unlisted technical disclosure for reviewers.

What search engines may index

Public study hubs, the ayah workspace, named tafsir verse queries, Learn Hub, topics, map, Seerah, and the teacher/parent landings are indexable. Signed-in teacher and parent panels, student rosters, invite codes, assessments, /about, and /methodology stay out of search.

FAQ

Will AI Quran invent verses?

No. The policy is that AI will never invent Quran verses, rewrite Quran text, invent hadith, or present an AI opinion as a fatwa.

Where is the full technical write-up?

The laboratory methodology page stays unlisted. This page is the public trust summary: retrieve-then-cite, abstention, and escalation to a local scholar.

Can I book a scholar here?

No. Live scholar booking is not available. Contact the team for product questions, not personal rulings.

Are teacher and parent dashboards public?

No. /teachers and /parents explain the product. /teacher and /parent panels, student progress, and assignments stay signed-in and out of search.

Cited answers are educational support, not a fatwa. Live teaching is not bookable.