AI was everywhere at Money20/20 — but where is it actually being used?

AI was one of the most visible themes at Money20/20 Middle East 2026, running through the show's programming and product pitches and through the announcements in Money20/20's press releases . That kind of near-universal presence at a fintech conference raises an obvious question: how much of it reflects AI actually running in production inside financial businesses, and how much is marketing language applied to features that are still early, experimental, or aspirational? The honest answer is a mix of both — but the deployed use cases cluster in a smaller, more specific set of places than the show floor's messaging might suggest. Why the question is worth asking at all Every major technology conference of the past several years has featured some version of this same tension: a wave of genuine underlying progress in a technology, paired with an even larger wave of vendors and exhibitors attaching that technology's name to products that may only use it in a limited or marketing-oriented way. Financial services conferences are particularly prone to this because AI's promise in finance is genuinely large — fraud losses, underwriting errors and compliance costs are all measured in real money, which gives every vendor a strong incentive to claim their product addresses one of those problems using AI, whether or not the underlying deployment matches the pitch. With AI messaging visible across exhibitors at Money20/20 Middle East 2026, the marketing-versus-deployment question is a natural one to ask of this year's show. Fraud detection: where AI has the longest track record Fraud detection is the AI use case with the deepest history in financial services generally, predating the current wave of generative AI enthusiasm by years. Pattern-matching and anomaly-detection models are well suited to spotting fraudulent transactions because fraud, definitionally, deviates from normal behaviour in ways that machine learning models can be trained to flag. That maturity makes fraud detection one of the more credible AI claims at a conference like this one — it is an area where "AI-powered" is more likely to describe genuinely deployed technology than a newly bolted-on feature. Fraud models also benefit from a structural advantage that newer AI use cases lack: financial institutions have been collecting labelled transaction data — confirmed fraud versus confirmed legitimate activity — for long enough that there is a large, well-understood training dataset to build on, which is a big part of why this category matured earlier than most others in financial AI. Underwriting and credit decisioning: the newest heavyweight use case The clearest example of AI in production from this year's announcements is Abwab.ai, whose seed round — covered in TechScoop's report on its $4 million raise led by Speedinvest — came with a specific, verifiable claim: more than SR10 billion in loans already processed through its AI underwriting infrastructure. That figure matters precisely because it is a usage metric tied to real loan volume, not a projection. Underwriting is a natural fit for AI's economic case: better risk assessment produces measurably fewer bad loans and faster approval times, both of which show up directly on a lender's bottom line, which gives underwriting-focused AI companies a comparatively easy story to tell with real numbers behind it. It is also a category where the Speedinvest-led investor syndicate backing Abwab.ai — its first Saudi investment, alongside MEVP — appears to have specifically valued the processing-volume evidence over a purely forward-looking product pitch, which is itself a signal that sophisticated investors are treating verified underwriting volume as the more credible marker of real AI deployment in this space. Customer service: widely marketed, harder to verify Customer-service automation — chatbots, virtual assistants, and AI-driven support tools — is likely the single most heavily marketed AI use case at any fintech conference, in the Gulf or elsewhere, in large part because it is the most visible to a booth visitor who can simply interact with a demo in real time. It is also, generally, the hardest category in which to distinguish genuine production deployment at scale from a well-rehearsed demo. TechScoop has not seen company-specific usage data for customer-service AI comparable to Abwab.ai's loan-processing figure, so TechScoop treats this category as visibly present at the show but harder to verify as deployed at meaningful scale based on the sourcing available. Compliance: a growing but quieter use case AI applied to regulatory compliance — automating parts of the due diligence, monitoring and reporting that financial institutions must perform — tends to attract less booth-floor attention than fraud or customer service, precisely because it is an internal, operational function rather than something a conference attendee can interact with directly. That lower visibility does not mean it is less real;