Ruya Bank puts AI agents into banking production via a private cloud and open models in Ajman

Ruya Bank, a digital Islamic community bank based in the emirate of Ajman and licensed by the Central Bank of the UAE, has deployed three operational AI agent use cases into production in partnership with Dubai-based AI software firm Magor. These deployments focus on the corporate onboarding workflow, the most labour-intensive and costly phase of onboarding small enterprises and entrepreneurs, where the systems read, verify, and analyse documents, preparing comprehensive summaries for each file before routing it to human review.

A strict separation between data processing and decision-making sits at the core of the implementation, with agent tasks restricted to verification and matching while final onboarding decisions remain exclusively with human officers.The architecture runs Magor's MagOneAI platform entirely within the bank's private cloud, using open-source large language models integrated directly into its technical infrastructure. This setup ensures sensitive customer data remains on-premises, satisfying regulatory mandates on data sovereignty and residency.

The rollout aligns with standards from the Central Bank of the UAE requiring board-level governance, system transparency, direct human oversight on high-impact financial decisions, and consumer protection. Established in 2024, Ruya Bank operates without legacy technical debt, giving it architectural agility unavailable to incumbents constrained by decades-old core systems, and allowing it to build AI tools directly into its foundational workflows at lower cost than retrospective modernization.

Relying on private cloud infrastructure and open-source models gives banks a clear compliance pathway with regulators, even if it requires accepting trade-offs in capability compared with proprietary cloud-hosted models.The roadmap for the partnership, framed as a long-term engagement rather than a standalone project, includes expansion into collections, compliance, credit underwriting support, customer analytics, personalisation, and an internal staff assistant dubbed RuyaAI. Magor claims to serve more than 80 enterprise clients and to have deployed over 60 production AI implementations across financial services, healthcare, and government, though those figures have not been independently audited.

While the rollout offers a practical template for bringing AI agents into licensed banking production, the announcement disclosed neither transaction volumes, accuracy rates, and error margins, nor the actual time saved by compliance reviewers or the total implementation cost. Neither party named the specific open-source models powering the system, data points that remain critical for assessing Arabic document and text processing performance before this deployment can be evaluated as a full case study.