In service of the Qur'an, the Arabic language, and the Muslim world — built for everyone learning Arabic and seeking to understand Islam.
Barq is an AI that speaks Arabic fluently and never invents — every fact it gives is drawn from a verified knowledge layer (the Qur'an, Tafsir, Hadith, poetry, and language), not from guessing. It is built to be a trustworthy reference that helps you understand Arabic, the Qur'an, and Islamic culture.
What Barq does that general LLMs can't:
Every verse, number, line of poetry, and root is retrieved from a verified knowledge layer — never generated from memory. General models guess; Barq looks it up at the source.
It doesn't just extract the root; it knows what the root means. Barq learned — from the Qur'an and language corpora — a meaning-space where related roots sit close together: it sees that «حجب» (concealment) neighbours «ستر · غطو · سجف», and that «كتب» is one family (كاتب · مكتوب · كتاب). So it finds synonyms across roots, links concepts, and grasps a word from its semantic neighbourhood — grounded in data, not guesswork.
The unique edge: general LLMs approximate this statistically and slip on morphology, rare, and Qur'anic words; Barq carries an explicit, meaning-measured root layer that no general model has.
Every answer traces to a real source (verse reference, poem attribution) — trustworthy and verifiable. Non-negotiable for education, publishing, and religious contexts.
The engine, knowledge, and model are fully owned (not a rented black box), can run privately, and the "truth layer" can be licensed to other systems.
The family isn't a set of parallel models — it's a system for raising and measuring one. At its heart is Barq Kid: a pure Arabic AI, built from scratch on the roots of the language. Around him stand his teachers and benchmarks, who guide his learning and measure how far he has come.
The child the whole family is built for: an Arabic model built from scratch on the roots and patterns of Arabic — not adapted from a foreign model, not bound to a tokenizer alien to Arabic. He is the purest expression of the sovereignty vision: Arabic by origin, fully owned, learning his language from its root. The family around him are his references, teachers, and benchmarks — leading his learning and measuring his growth until he comes of age.
His Teachers & Benchmarks
A general Arabic assistant grounded in Barq's knowledge — gives the Kid the breadth of the language and a benchmark for general ability.
The reference for language, roots, and the Qur'an — the gold standard against which the Kid's mastery is measured.
Poetry & literature from a large, attributed corpus — teaches the Kid the eloquence and music of Arabic.
Deep, verified research across Barq's sources — teaches the Kid to ground every claim and never invent.
A planned expansion of the family: Linah (لينة) — Barq's sister, a grounded Arabic assistant specialized in Islamic jurisprudence (Fiqh), researching the fiqh of women and family; and alongside her Little Linah (لينة الصغيرة) — her pure, from-scratch Arabic model, born the same way as Barq Kid.
A grounded Arabic assistant specialized in Fiqh, researching the jurisprudence of women and family — with the same principle that defines Barq: never invents; every question tied to its source in the Qur'an, the Sunnah, and the words of scholars — presenting positions, not issuing fatwas.
Linah's pure root-based model — learning Arabic from its root just as Barq Kid does; the family is built for her and her brother alike.
Inventor Abdullah Nabil Al-Khatib's portfolio includes one granted U.S. patent and three pending Saudi applications. Each link opens the public illustrated report for that invention.
Run-time business application development. Illustrated report →
AI-enhanced development with deterministic language-model behavior and layered compliance.
Multi-policy textual memory that preserves original text and validates citations.
Alignment of readings with manuscript images, visual evidence, and candidate references.
The links open the Arabic public report and drawings; application statuses follow the owner's records.
R&D & Enterprise Platform Architect, inventor, and CEO — 25+ years building bilingual (Arabic/English) enterprise systems, with one granted U.S. patent and three pending Saudi applications. The architect behind Barq's sovereign, trustworthy Arabic AI.
One granted U.S. patent and three pending Saudi applications. View the public report →
CEO and R&D Director at Visual Software Company (visualsoft.com) — leading platform strategy, innovation, and digital transformation.