Khabir launches Sovereign Data Initiative for GCC AI programmes
The Initiative provides dedicated Arabic-native data operations to national AI programmes, with full jurisdictional data residency and confidential MoU.
Read announcementKhabir is the authority on Arabic-language human data for sovereign AI programmes, frontier laboratories, and Gulf enterprises. We vet domain experts across the GCC and Levant — linguists, scholars, jurists, physicians, engineers, and operators — and convert their judgment into the highest-grade post-training data, evaluations, and reinforcement-learning environments.
Khabir is now accepting expressions of interest from sovereign AI programmes seeking dedicated Arabic-native data operations. Confidential briefings available with Memorandum of Understanding.
Every Khabir service is delivered by the same vetted network, against the same calibration rubric, with the same audit standard expected by frontier-laboratory and government procurement.
Arabic-native SFT, RLHF, and DPO datasets across MSA and five regional dialects. Instruction-following, reasoning chains, and preference data authored by domain experts.
View specificationsCustom benchmarks and continuous evaluation programmes for Arabic reasoning, Sharia compliance, GCC legal accuracy, dialect fidelity, and domain-specific Q&A.
View specificationsDomain-specific virtual workspaces for training and evaluating agents on real GCC workflows — banking operations, government services, clinical triage, contract review.
View specificationsEgocentric demonstrators across the GCC and Levant performing standardised tasks — household, retail, hospitality, light industrial — for humanoid pre-training.
View specificationsSubscription access to vetted Arabic-speaking specialists for ad-hoc human-in-the-loop work — Sharia scholars, GCC jurists, native dialect linguists, clinical experts.
View specificationsLong-term, jurisdiction-resident data operations for national AI initiatives. Includes data residency, audit trails, and dedicated Arabic-speaking programme management.
View specifications"The only way frontier models continue to learn is through net new human data. Owning the Arabic-speaking expert layer is the difference between a sovereign AI strategy and a translation layer."
— KHABIR FOUNDING THESIS, 2026Khabir is established in Dubai with a mandate to operate as the regional standard for Arabic-native human-data services. The organisation supplies sovereign AI programmes, frontier laboratories, and major Gulf enterprises with calibrated expert intelligence — the resource that frontier models cannot produce on their own.
Our standards are calibrated to frontier-laboratory procurement requirements. Every contributor on the Khabir network passes a written and verbal vetting before they receive a single contract.
The Initiative provides dedicated Arabic-native data operations to national AI programmes, with full jurisdictional data residency and confidential MoU.
Read announcementThe Standard codifies five-axis expert calibration — domain depth, uncertainty handling, reasoning transparency, communication, and trust integrity.
Read announcementAn evaluation programme covering eight fiqh categories and four schools of jurisprudence, authored by a panel of credentialed scholars.
Read announcementDiscovery briefings are 30 minutes, conducted under non-disclosure, and produce a written scoping memorandum within 48 hours. Available in person across the GCC, or via secure video conference.