AI-Native AML: LLM Detection, Explainable Decisions & Human Oversight | Flagright, B4B Payments, Zero | The Fintech Show #158

Baran Ozkan, Flagright, Luke McAlpine, B4B Payments, and Richard Theo, Zero, center on how AI-native AML, agentic workflows, and tighter KYC/monitoring/card partnerships are reshaping compliance amid fast-evolving social-media-driven fraud, and how natural-language interfaces and explainability keep humans in the loop.

AI is now essential in financial services: criminals are using it, and firms need it for efficiency. Ozkan argues large language models can automate costly AML work and cut repetitive tasks, shifting operations to “AI-native” with machines doing most of the work and humans providing oversight and auditability.

Theo brings the consumer fintech reality. Fraud morphs fast, money-mule recruitment among young people and scams like fake concert tickets are spreading on social media and notes a policy gap: while the FCA heavily regulates finance, most scams originate on lightly accountable platforms. On controls, Zero layers strong KYC (documents, biometrics, address, sanctions/PEP, UK fraud-database checks) with ongoing monitoring: Flagright for bank-to-bank transaction monitoring and Marqeta for card controls, combining rules and AI to spot abnormal patterns and act quickly.

Ozkan outlines Flagright’s path: Y Combinator credibility, trust signals (ISO/SOC 2), and a clear thesis: AI-native AML. The product grew from risk scoring, monitoring and case management into sanctions screening, AI agents, ontology and L3 QA. “AI Forensics” agents suppress screening false positives and compress multi-minute analyst workflows into seconds via conversation. The aim isn’t to replace people, but to give them leverage.

McAlpine focuses on scale and changing engagement models. B4B Payments is exploring natural-language interactions, agentic payment flows, and watching the Model Context Protocol, while still improving today’s APIs and developer experience. The team uses AI coding/prototyping tools where they add value. On compliance, B4B works with Flagright on advanced pattern matching, AI-assisted forensics and LLM-drafted outputs (e.g., SARs), plus horizon-scanning for regulation and even codebase impacts. Explainability is a priority; decisions must be traceable, with humans in the loop at a larger scale.

Partnerships and execution speed stand out as Ozkan stresses “show, don’t tell” and strong support, buyers call Flagright intuitive, with one MLRO deploying it at three firms. McAlpine echoes the collaborative roadmap.

Looking ahead, Ozkan plans to double down in core markets such as North America, the UK/EU, APAC and the UAE; with new hubs in London and New York and a push on brand awareness. Theo’s growth plan recognizes UK “free banking” economics: Zero isn’t a bank but an EMI distributor, so new revenue features are coming, “planet-safe” instant-access savings, carbon-offsetting tied to a green score, ESG investments, sustainable-spend rewards, and a community-share model reserving 20% of equity for customers.

The bottom line is that compliance is moving from rules plus manual review to AI-native systems with human oversight; client engagement may shift from screens to natural language and agents; and tight partner stacks across KYC, monitoring and card rails are becoming the practical backbone.