The event was conceptualised, planned and developed by Mr Aaron Ang, Chief Commercial Officer of CWG, who leads the RealRelay project. Mr Ang built the day around one challenge. Participants were handed a simulated fraud case and asked to use agentic AI to find out what had happened. Within hours, they had worked through tens of thousands of transaction records, traced the money, flagged the entries that did not add up, and shown how they got there. The same job, done by hand, would take an audit team weeks. The event was graced by Mr Ong Kah Kuang, Assistant Chief Executive of the National Youth Council.
That is what RealRelay Fintech sets out to do. Work that is slow, manual, and easy to get wrong, such as reconciling accounts, checking documents, and finding the one record in a million that should not be there, becomes something an agent on the platform does accurately, in hours, with every step on record. Fraud detection is the clearest example, and it is where the fintech vertical begins. In finance, though, a fast answer counts for nothing unless the institution can trust it. Finance is the sector where trust in AI matters most, and Singapore has moved first. On 7 October, the Monetary Authority of Singapore (MAS) issued its Guidelines on AI Risk Management.
They cover every financial institution and every form of AI, agents included. Boards are expected to oversee AI risk, keep an inventory of each AI use, manage third-party AI, and control the whole life cycle, from testing to monitoring. The guidelines take effect in October 2027. In August, MAS also committed S$220 million under FSTI 4.0 to get AI into the sector faster, including an AI Pathfinder track for solutions listed on PathFin.ai. Every agent on RealRelay passes through the Trust and Control Layer. The platform’s own security agent tests it before it goes live and tests it again each time it changes. Permissions, logging, retention and access are controlled centrally, and every agent carries a risk rating that covers the data it can touch, the actions it can take and how it behaved under test. That rating, the agent inventory and the full log are the same things MAS now asks financial institutions to keep. The Trust and Control Layer draws on Edvance’s more than 20 years in cybersecurity.
“Singapore has been clear about how it wants AI to enter finance. Innovate, but with trust and control in place from the start. We believe that is the next phase of fintech, and we built RealRelay Fintech on exactly that principle. The Trust and Control Layer is the foundation of the platform. Every agent is tested before it goes live, and the institution keeps full control over what it can see and do. When people can see that, they stop hesitating. That is how banks, funds and founders will actually put AI to work, and that is what we are here to make possible.”
-Mr Ang
What RealRelay Fintech does
At launch the vertical covers four areas of work.
Research and analysis. Agents that read filings, news and market data, run scenarios and produce forecasts with the reasoning shown.
Decision documents. Credit memos, investment committee packs, portfolio risk briefs and client review notes, drafted from a firm’s own data and templates.
Fraud, compliance and document checks. Agents that reconcile transactions, flag anomalies, check onboarding documents against policy and keep a record of every check.
Learning. A guided track for students and new joiners, using the same tools professionals use, at a price a student can afford.
Institutional deployments, including Singapore-hosted and on-premises options, are available on request.
RealRelay has two layers sitting on a Trust and Control Layer. The Token Exchange launched on 30 September 2026 and puts more than 300 mainstream AI models behind one subscription, routing each task to a suitable model at a lower cost than buying them separately. The Agentic Layer sits on top. Users describe what they need in plain language, assemble an agent from ready-made modules, test it and deploy it in one click, with no code. Three verticals run on these layers: education, cybersecurity and now fintech.
Education came first, and it is where the platform found its first users. At the APEX Bootcamp in September, students and young working adults with no technical background built working agents in two days. Interest from educational institutions has followed, and CWG is finalising details with these institutions to provide RealRelay and the full AI ecosystem around it to their students. Fintech now follows the same path, on the same layers and under the same controls.
