Mashreq Bank has significantly reduced the time required for system integration by deploying artificial intelligence agents, according to a Monday announcement from IBM. The Dubai-based financial institution reported that the pilot program cut integration design and troubleshooting time by up to 70%. Tasks that previously required two to three hours of manual effort are now completed in minutes.
The bank is utilizing IBM Bob, an AI software engineering platform, to handle complex wiring between its digital banking channels, payment systems, and enterprise infrastructure. This automation addresses a common industry bottleneck where every new payment rail, app feature, or partner connection must be manually wired into existing systems. The cleanup process when these connections break is also typically performed by hand.
Mashreq AI Agents Handle 4,000 Monthly Requests
To support this workflow, Mashreq built a secure gateway based on the Model Context Protocol (MCP). This infrastructure allows AI agents to access integration knowledge, application programming interface specifications, architecture standards, and operational context without compromising security. The bank currently processes approximately 4,000 AI-agent requests per month through this gateway.
The AI tools are integrated across the entire engineering cycle. Engineers use the agents for solution design, planning, development, code review, and production troubleshooting. Santosh Vaidya, Mashreq’s head of payments and digital services, stated that the bank aims to create an engineering environment where AI agents work alongside human teams.
Mohamed Abdel Razek, Mashreq’s group head of technology, transformation, and information, emphasized the need for speed without sacrificing quality. He noted that the bank’s integration platform averages 650 transactions per second across digital channels, payments, and enterprise systems. The investment in these technologies is intended to help engineering teams move faster while maintaining the resilience expected by customers.
Industry Bottlenecks in Real-Time Payments
Mashreq’s experience highlights a widespread challenge in the financial sector. A PYMNTS Intelligence report found that 53% of financial institutions identify manual-intensive internal processes as their primary obstacle to modernizing real-time payments. Another 52% cited outdated technology, while 47% pointed to siloed payment systems.
Data quality and fragmentation also pose significant hurdles. Three in ten financial services leaders at U.S. companies with at least $1 billion in revenue named data issues as their main constraint on AI adoption. AI agents offer a potential solution by accessing core banking systems through the same APIs used by consumer apps or through MCP connections that expose approved functions as tools.
This shift could alter the economics for systems integrators such as Accenture and Capgemini, as coding agents begin to automate parts of mapping, coding, testing, and documentation. Mashreq’s pilot builds on earlier upgrades, including the 2024 connection of its corporate online banking channels to Aani, the UAE central bank’s instant payments platform, which enabled round-the-clock transfers.
The integration platform rebuild involves IBM and Gulf Business Machines, running on IBM integration software, Red Hat OpenShift, and Confluent’s data streaming technology. Mashreq is also exploring an on-premises deployment of IBM Bob for critical work that requires locally hosted models to meet data residency and regulatory requirements.
Source: PYMNTS
