The Salsabeel vision: documented knowledge, models we develop and own, and institutional work that builds lasting capability for future generations.
Vision updated: 13 October 2026
We began with the resources we had and continue with determination. Our ambition is for the efforts of the Arab and Islamic world to come together in building capability worthy of its language, the depth of its civilization, and the aspirations of its generations—through creative minds, established institutions, and major investment. This work has room for every serious contribution. Its horizon extends beyond the success of any individual or company: an original contribution to the development of AI, knowledge that benefits the world, and an achievement on which future generations can build.
Serving Arabic, Islamic knowledge, and cultural heritage is a purpose whose value we do not make conditional on short-term financial returns. We build on global scientific progress and contribute knowledge and tools that others can examine and develop further.
Privacy, data location, and usage rights depend on the institution and deployment. We work toward clear deployment and control options, with external dependencies disclosed.
Arabic morphology, Qur'anic quotations, and hadith attribution have specific requirements. We connect language capabilities with specialist tools and sources that can be reviewed.
Model size and tools should fit the task, cost, and required accuracy. Performance claims need documented samples and defined comparisons; size alone does not establish them.
Models we develop, organized data, evaluation standards, and research expertise work together to serve education, research, and institutions.
Barq Fata is an Arabic model we own and develop, with approximately 1.6 billion parameters. It operates within a system that includes retrieval, linguistic tools, and verification. Our evaluations therefore distinguish the model acting alone from the complete system.
We retrieve source texts, link answers to evidence, and test quotation accuracy and abstention when evidence is insufficient. These measures aim to reduce errors; they do not guarantee their absence.
Root and derivation analysis, Qur'anic adjacency research, and NMN structured textual memory each have their own scope, examples, and maturity stage.
Owning the Barq model does not mean owning every source or component. We distinguish our assets from third-party materials and respect their licenses and permissions.
Larger models support some escalation and improvement paths. We do not attribute results from those paths to Barq acting alone.
For heritage, we distinguish the atlas of manuscript catalogue records and repository links from research into reading manuscripts and documenting readings with visual evidence.
Founder and owner of Visualsoft and Salsabeel, leading research vision and architecture with more than 25 years of experience in enterprise platforms, Arabic–English systems, and AI-assisted development.
United States — US 8,782,095 B2 (Granted): Run-time enterprise application development. Illustrated report →
SA 1020261862 (Pending): Multi-region AI-enhanced enterprise application development, deterministic language-model behavior and layered compliance.
SA 1020268774 (Pending): Citation control using multi-policy textual memory.
SA 1020268792 (Pending): Manuscript-reading documentation aligned with visual evidence.
Our proposed revenue model covers institutional customization, source integration, deployment and support, and licensing capabilities we have the right to license. We are developing these paths to sustain research and public service. This is a business plan, not a statement of earned revenue or completed contracts.
The commercial priority is a clearly scoped institutional research and knowledge service, tested with users and assessed for practical value. Public knowledge access continues within available resources and policies.
Mission and sustainability details in the roadmap (Arabic) →
We are working toward Barq 7B with our own financial and technical resources; release approval depends on passing its evaluation.
21B and 51B are research horizons subject to evidence and feasibility. Expansion on this scale requires scientific and institutional collaboration, specialist expertise, and substantial long-term investment, with clear deliverables, responsibilities, and success criteria.
This work has room for scientific, technical, and knowledge contributions, and for serious institutional partnerships that turn accumulated capability into wider impact.
📧 Abdullah@salsabeel.ai · 📱 +966 50 496 3570 · 🌐 salsabeel.ai