About — Divine Message Academy

Divine Message Research is a closed-circuit research environment paired with a peer-reviewed scholarly journal. It gives structured access to primary sources across the full arc of scriptural history — from the Qur'an and classical tafsīr literature, through the Torah and the Bible, to the Dead Sea Scrolls — all in one place, for serious comparative and textual research.

Research focus

Research here is focused on the commons among the divine-message sources — the Qur'an, the Torah (Tawrah), the Bible, and the Dead Sea Scrolls: the themes, figures, narratives, and teachings these sources share. It is a call for work on what they hold in common, not on every topic within them.

Access to the research environment is by invitation, extended to motivated researchers and the scholar community.

What this is

This is a personal research tool for comparative study of scripture and revelation across traditions — what began as a tool for studying the Qur'an alongside the classical Islamic scholarly tradition that surrounds it has grown into something wider. Its focus is the commons among four primary sources: the Qur'an, the Torah, the Bible, and the Dead Sea Scrolls.

The Qur'an is indexed in its own right, with the major hadith collections and roughly forty classical Arabic works of tafsir, hadith commentary, and philology as supporting literature. The platform also includes the Hebrew Bible (Torah/Tanakh — Westminster Leningrad Codex, the standard Masoretic text, original language), the World English Bible (a modern English translation made directly from the original Hebrew and Greek, covering both Old and New Testament), and the Dead Sea Scrolls (Martin Abegg, James Bowley, and Edward M. Cook's transcription, via the ETCBC — both the biblical scrolls and the wider Qumran community's own non-biblical writings, licensed CC BY-NC 4.0, non-commercial use only), each indexed as its own dedicated category. A separate library of general academic textbooks (unrelated to religious studies) is also indexed, kept apart from the religious corpus. Beyond the platform's small default core, sources are reached by selecting them directly.

What it does

Sources Only mode — pure retrieval, no AI generation at all. Hybrid keyword + vector search with cross-encoder reranking returns the actual passages, verbatim, with citations.
AI Research mode — the same retrieval feeds a grounded synthesis step. The AI is instructed to answer only from the passages it retrieved and must cite them; it is not meant to draw on its own background knowledge of Islamic topics.
You choose exactly which sources to search, or let the AI pick based on relevance to your question.
Adjustable depth, detail, and academic register, and a cross-lingual bridge so a question in one language can still retrieve source material in another.
Every answer is traceable back to a specific passage in a specific, named source.

What it does not do

It is not a mufti, a scholar, or a fatwa engine, and it issues no religious rulings.
In AI mode, it does not draw on the model's general or background knowledge of Islam — only on the specific passages retrieved for that question. That is a deliberate design choice, not a limitation of the underlying AI model.
It does not certify hadith authenticity grading, scholarly consensus, or completeness of coverage — the corpus, however large, is necessarily partial.
It is not a substitute for studying with a qualified teacher, and nothing here should be treated as the final word on a matter of belief or practice.

How to use this platform

Getting started

Ask a question — type it in the question box and press "Ask" (Enter also works; Shift+Enter starts a new line).
Confirm how it was understood — before the real search runs, the platform shows "I understood your question as: …" so you can catch a misreading early. Choose "Search this" to proceed, "Edit question" to rephrase, or "Search as typed" to skip the rewrite entirely. Check "Don't ask again — search immediately" to stop seeing this step.

Research modes

🔬Findings — pure retrieval, no AI-written text at all. Returns the actual matching passages verbatim, ranked by relevance, each with its own citation.
🤖AI Synthesis — the same retrieval feeds a grounded synthesis step: the AI is instructed to answer only from the passages it retrieved, and must cite every claim.
🔍Exact Search — bypasses relevance ranking entirely and finds every literal occurrence of a word or phrase across the library (e.g. "how many verses contain the word Allah"). Type the exact word or phrase itself, not a full sentence.

Choosing sources

Manual selection — open the source list and check exactly the works you want searched. Selecting sources yourself restricts the search to only those sources.
🍀Let AI choose (Findings and AI Synthesis) — searches a curated core set (Qur'an, tafsir, hadith, biography) and narrows to whichever works actually had something relevant, so you don't have to guess. It never reaches outside that curated set — specialized sources (dictionaries, poetry, modern science, Bible/Torah/Dead Sea Scrolls, etc.) still need to be picked by hand.
🌐Term Assist (Exact Search only) — makes one AI call to predict the same term's natural Turkish/Arabic/English/Hebrew equivalent, then searches all four exact forms at once, so a source that never spells it the way you typed it is still found. The exact terms searched, and how many matches came from each, are shown afterward.

Depth

The three colored dots next to the language switcher set how many passages are pulled from the library before an answer is written: green = Normal, blue = Comprehensive, grey = Unlimited. Normal is fastest; Unlimited is the most thorough and slowest.

Core, Quotes & Translate Quotes (AI Synthesis only)

Core — switches the writing to a terser, more advanced register and skips explaining everyday terms. It never changes which sources are searched or how grounded the answer is.
📜Quotes — shows the exact excerpt text each footnote cites, right below the sentence that cites it, so you can check a claim against its real source without leaving the answer. Only works with Footnote-style citations.
🌐Translate Quotes — once Quotes is on, adds an AI translation of each excerpt into your current interface language, right underneath the original text (never replacing it) — useful when an excerpt is in Hebrew, Arabic, or another language you may not read directly.

Citations & references

Every answer or finding is traceable to a specific passage in a specific, named source — a full reference list is shown below every answer.

The default citation style is Footnote (numbered, in-text). APA, Chicago, and Islamic Studies styles are also available — see Advanced Mode below.

Listening & exporting

🔊Listen — reads the current answer aloud; use Pause/Resume mid-playback and the Slow/Normal/Fast selector to change pace. Several voice engines are available per language, chosen from the dropdown next to Listen.
⬇️Download Audio (MP3) — saves the narration as a single MP3 file.
📄PDF / Save — opens your browser's print dialog for the current answer, which you can print or save as a PDF from there.
⚖️Compare — planned but not yet available: seeing the same question answered at different depths side by side.

Language & interface

Switch the interface between English, Turkish, Arabic, and French at any time from the top of the page — Arabic automatically mirrors the layout right-to-left.

A cross-lingual retrieval bridge means a question asked in one language can still surface source material written in another.

Accounts & My Notes

Register — with Google, or an email and password (email accounts require verifying your address before you can log in).
📝My Notes — once logged in, opens your own private notes folder in a new tab, where you can write and format your own research notes, visible only to you.

Advanced / Pro mode

Typing /research/mod or /research/pro directly in the address bar forces every request to use a higher AI-quality tier (Moderate or Pro) and reveals two extra manual controls: Citation Style (Footnote, APA, Chicago, or Islamic Studies) and AI Quality itself, for overriding the tier per request.

Higher tiers use a stronger underlying model at more pipeline steps — more accurate, but slower.

How publishing works

The Journal is paired directly with the research environment, so a paper can move from notes to print without leaving the platform.

1.Research and compose — you research and write up your findings directly in the platform.
2.Draft — the “Make This a Scholarly Article” tool takes only your own notes and formats them into a properly structured, closed-circuit paper.
3.Confirm — you review the formatted draft yourself and confirm it before anything moves forward.
4.Submit — you submit it to the Journal directly from the platform.
5.Review — the Journal's editors and peer reviewers review the paper and share their feedback with you.
6.Publish — once ready, your work is published.

Every step above is free of charge.

Ways to take part

Access to the research environment is by invitation, extended to motivated researchers and the scholar community. Scholars are invited to take part in any of the roles below.

Reviewer — evaluate submitted papers within your area of expertise.
Author — publish your own research through the platform and journal.
Committee Member — help shape the platform's scope, standards, and direction.
Editor — oversee submissions and the review process for the journal.

To request an invitation or to take part, get in touch and say which role interests you, along with anything you'd like to add — questions, areas of interest, or availability. Everything described here is free of charge.

Get in touch

Strengths & opportunities

Cross-references dozens of classical tafsir and hadith works in a single search — work that would otherwise take hours by hand.
Makes rarely-digitized, often untranslated classical Arabic works searchable and directly citable.
A multilingual interface and cross-lingual retrieval lower the barrier for readers studying in a language other than Arabic.
The no-AI retrieval mode exists specifically for when even a grounded AI summary isn't wanted — just the sources, as written.

Limitations & risks

The corpus is fixed and finite. Silence on a topic here does not mean silence in the wider scriptural and scholarly tradition.
The underlying texts may carry translation, OCR, or digitization artifacts from how they were originally sourced.
Like any AI system, the AI Research mode carries some residual risk of misreading nuance or making an error, even with grounding and citation enforced — always check the cited passage itself, not just the AI's summary of it.
This is an actively-developed personal project, not a maintained academic product. Expect it to change, and verify anything consequential independently.

Why this exists

I built this platform originally for my own study of the Qur'an and the classical commentary tradition around it — that study grew into a wider interest in comparative study of how different traditions describe divine revelation, and the platform grew with it. I'm making it available to invited researchers and the wider scholar community, who may find it useful for similar study, with the limitations above kept in mind.

Open Divine Message Research