Afshin Lamei

What you will learn:

Scaling manual threat modeling is a significant challenge. At Mollie, microservices and payment flows evolve daily. This forces security engineers to choose: slow product development by attending every architecture review, or risk critical design flaws by stepping back. We addressed this bottleneck by building an AI security assistant. This assistant acts as a first-pass peer reviewer for threat models and system design documents. Our approach does not aim for full automation. Instead, the LLM assistant gathers initial context, flags missing security controls, and highlights policy gaps. It then hands off a refined analysis to human security engineers for final triage.