Citation-anchored research
for the Islamic sciences.
Every citation traceable. Every term preserved. Built by ʿulamāʾ, for the people inside the tradition: ʿulamāʾ, muftīs, advanced students, and serious researchers who can finally trust an AI as a partner in their work.
This is not a pitch deck.
Istidlal is already running.
For the past month, Istidlal has been in active testing with a small circle of ʿulamāʾ and iftāʾ graduates — producing real citations against 815,000+ indexed passages across the corpus you'll read about below.
The engine is built. The retrieval pipeline is shipping. The terminology guardrails are catching mistakes that generic AI doesn't even know are mistakes. Every answer in front of a tester sharpens the next one.
While you read this, the bot is answering questions.
Generic AI fabricates fiqh.
ChatGPT, Claude, Gemini. Every one of them confabulates Islamic citations. They invent ḥadīths. They misattribute fatwas. They cite books that never existed. They don't know they're wrong, because they were never trained to be right.
"It is mentioned in Radd al-Muhtār vol. 4 p. 217 that women…"
"Per Radd al-Muḥtār 6/348, the ruling is…"1
The most comprehensive Hanafi research base, anywhere.
Centuries of fatwas, the foundational mutūn, the trusted dār al-iftāʾ rulings, and classical ḥadīth, all indexed, embedded, and searchable as a single body of evidence.
New sources added daily, the corpus is not limited to these.
Built by scholars
with engineering expertise —
for the people inside the tradition.
Every retrieval pipeline, every prompt rule, every terminology guardrail — designed by people who have sat through years of ʿilm and know what a real citation has to look like before it can survive a question from a discerning teacher.
An ʿālim and graduate of a major Darul Uloom in the Hanafi tradition. Multiple takhasṣuṣāt already in hand, Iftāʾ and Adab — and currently undertaking a takhasṣuṣ in Hadīth. Active in the community as an Imam and ongoing student of the sciences.
In parallel: a senior software engineer with formal degrees in both Software Engineering (MS) and Cybersecurity (BS), and 5+ years of professional experience shipping real systems, full-stack web applications, AI-driven chatbots, automation pipelines, multi-channel platforms, secure backend architectures, and IoT integrations. Recent focus: applied AI for production knowledge bases.
Trained by ʿulamāʾ and mashāyikh of the tradition, fluent in the classical texts, and fluent in the engineering disciplines that build serious software. Not a side project run by an outsider learning ʿilm on the fly, nor a software project run by ʿulamāʾ outsourcing the technical work and hoping for the best. It is the first time both qualifications have lived inside the same person on a project of this scope. That is why this exists when nothing else does.
Identity held back during pre-launch. Will be public at the launch announcement.
Reclaiming the golden age.
For five hundred years, Muslims sat at the frontier of human knowledge. Al-Khwārizmī gave the world algebra. Ibn al-Haytham gave it the scientific method. Ibn Sīnā set the template for medical reasoning that taught Europe for centuries.
The intellectual culture that produced these breakthroughs was not separate from ʿilm. It was ʿilm, applied, systematized, and shared.
Modern AI is the most consequential information technology since the printing press. It is only fitting that ahl al-ʿilm shape how it serves their fields, instead of leaving it to whoever gets there first.
Istidlal is a small piece of that work. A research tool, designed by people inside the tradition, for people who carry it forward.
What is stopping us from using AI
to its fullest potential?
Tens of thousands of makhṭūṭāt, manuscripts of fiqh, tafsīr, ḥadīth, and ʿaqīda, sit in libraries from Istanbul to Patna, untranscribed and effectively inaccessible. Books that scholars dismiss as "lost" are not lost. They are stacked in cataloguing rooms, available only to whoever happens to be physically present and trained to read centuries-old script.
Critical works are fragmented across languages. A scholar researching a modern question may need to triangulate Arabic from Radd al-Muḥtār, Urdu from Maḥmūdiyya, and a contemporary Turkish risāla. Most never attempt the cross-language synthesis. The evidence is there, the time is not.
And even when a source is on the shelf, verifying a single quote takes time, different editions have different pagination, chapters get re-numbered, and tracking down the right copy can mean hours in a library. The evidence is there, but the cost of checking it carefully is high enough that careful checking has quietly become a luxury.
What if a serious researcher could query all of this in seconds?
Hanafi fiqh first.
The whole tradition next.
Hanafi fiqh is the launch case because nothing of this depth exists for it yet. The architecture is built once; the corpus expands forever.
One platform, one citation discipline, one growing library, not a half-dozen siloed apps that never talk to each other.
Not for everyone, and that's
by design.
A research tool that can surface citations on every fiqh question is exactly the kind of tool that, in the wrong hands, accelerates the issuance of poorly-grounded rulings by people without the training to weigh evidence properly. We gate it intentionally.
Every query costs us money, embedding API calls, retrieval, the LLM that writes the answer, the server that holds 800,000+ indexed passages, the ongoing engineering. Subscriptions are how we keep the project alive without selling out to advertisers, mining your queries, or training a competitor on your research.
You pay for tools you respect. This is one of those.
What scholars are saying.
On a recent question involving contemporary financial transactions, I had three Urdu sources, two Arabic mutūn, and a recent darul iftāʾ ruling that all needed to be cross-referenced. Istidlal surfaced the relevant passages from each in under a minute, with page numbers and direct links to the original text. What used to take me a full evening took fifteen minutes. Every passage I clicked through to was exactly where it claimed to be, nothing fabricated.
What sets it apart from generic AI is restraint. ChatGPT will silently write ḥarām when the source says impermissible, a real and consequential distinction in our field. Istidlal preserves the exact term every time. When two of the sources I queried disagreed on a ruling, the bot presented both positions with full attribution and let me weigh them. That is the discipline I want from a research tool: surface the evidence, leave the verdict to the muftī.
I tested it on questions I already knew the answer to, to see if it would hallucinate. It didn't. On a question about ḥukm al-imāmah for someone with a particular condition, it pulled passages from Radd al-Muḥtār, al-Baḥr al-Rāʾiq, and Tabyīn al-Ḥaqāʾiq, all with verifiable page references that I cross-checked against my print copies. Three of three were exact. For work that depends on rigorous citation, this is the first AI tool I trust with serious research.
Honest answers.
Istidlal is a research tool.
It does not issue rulings.
- ◆Surface relevant fatwas across decades of dār al-iftāʾ archives
- ◆Cite classical mutūn with verifiable book + page references
- ◆Preserve fiqh terminology with the precision scholars expect
- ◆Cross-reference Urdu, Arabic, and English sources in one query
- ◆Hand the researcher the raw evidence, they synthesize and rule
- ◇Issue fatwas
- ◇Replace your training, ijāza, or judgment
- ◇Make rulings on its own evidence-weighing
- ◇Substitute for verifying a citation against the original print
- ◇Pretend to know things outside its corpus
Be there when we open.
First access goes to ʿulamāʾ, muftīs, and iftāʾ graduates. No spam, no follow-ups, no newsletter, just one email when we open the doors.