Engineering Reliable AI: From Automotive Compliance to Multi-Agent System Quality

Tokyo AI (TAI) hosts a technical evening on what it takes to move AI from experimental proofs of concept to reliable, production-grade enterprise systems. The session spans three layers of the problem: safety-compliant AI governance under strict regulatory frameworks such as the automotive industry, system-level quality assurance for multi-step agentic workflows, and the post-deployment human and behavioral factors that decide whether a technically sound system is actually adopted. Three talks build from macro to micro and back to operations. Alireza "Andrew" Sharifikia (Corpy & Co.) maps a practical roadmap for enterprise-grade AI compliance in the Japanese automotive industry. Raphael Aubel (Corpy & Co.) extends QA4AI's System Quality axis to agentic systems, covering how to evaluate full agent trajectories rather than isolated responses. Tomomi Tanaka (founder, Behavioral AI Lab) examines the missing human layer behind post-deployment failures, including trust, incentives, and user adaptation. The format runs from 18:00 doors to 21:00 close, with a networking session after the talks. It is aimed at QA engineers, AI engineers, researchers, and technical teams building or evaluating reliable production AI in enterprise settings.