How does a fast-moving fintech innovate with AI while staying fully compliant in one of Europe’s most regulated industries? Ryan Chaves, head of machine learning and AI at Mollie, sat down with Techzine TV at Google Cloud AI Live in Amsterdam to share how his team is blending ML with modern LLMs to build real products that serve both internal teams and customers.
Ryan explains how Mollie runs more than a dozen classical machine learning models in production for risk and fraud, while layering AI on top for new use cases, from an AI-powered customer support assistant to a bill-scanning feature that outperforms traditional OCR. He also dives into how Mollie maintains tight relationships with Dutch regulators like the DNB and the FIU, and how compliance with GDPR, the EU AI Act, and EU data residency requirements actually sharpens AI practice rather than limiting it.
On the technology side, Ryan walks through Mollie’s journey on Google Cloud Platform, from building its ML platform on Vertex AI (now the Agent Platform) to developing its own LLM gateway from open source before Google offered a managed version. He reflects on the balance between in-house innovation and adopting managed services, the organizational challenges of AI adoption, and how tools like Claude Code are fundamentally changing what it means to be a software engineer today.
🔑 Key takeaways:
• Classical ML and generative AI are complementary, not competing technologies
• AI is used both internally for efficiency and externally in customer-facing Mollie app features
• Regulatory compliance (GDPR, EU AI Act, DNB, FIU) is treated as a driver of good AI practice
• Mollie built its own LLM gateway before Google offered one as a managed service
• The entire tech stack runs on GCP, with Vertex AI / Agent Platform as the ML and AI foundation
• Hiring technically creative people who can collaborate is more important than pure technical skill
• AI is reshaping the daily role of engineers: less manual coding, more guiding and planning
• Swapping out LLMs as they improve gives Mollie a near-automatic upgrade path
⏱️ Chapters:
0:05 – Introduction: Ryan Chaves, head of ML & AI at Mollie
0:29 – Mollie: from payments to full fintech platform
1:24 – Classical ML in production: risk and fraud
1:55 – Fitting AI into a regulated payments business
3:02 – AI features: customer support and bill scanning
4:23 – Navigating the EU AI Act, GDPR and regulators
6:30 – Building on Google Cloud: GCP, Vertex AI and Agent Platform
7:27 – Building an LLM gateway: staying ahead of big tech
10:06 – AI adoption: organizational change beats technical hurdles
12:49 – How AI is reshaping the role of engineers
👉 Visit techzine.eu for more in-depth articles and analysis.
📺 Keep watching on techzine.tv, subscribe and share your favorite videos with colleagues.