​Description

​This Tokyo AI (TAI) community event focuses on the engineering and operational challenges of transitioning artificial intelligence from experimental proofs of concept (PoCs) to reliable, production-grade enterprise deployments. Bringing together technical experts, researchers, and systems architects, the session addresses three critical dimensions of modern AI systems:

• ​Navigating safety-compliant AI governance within strict regulatory frameworks like the automotive industry.

• ​Establishing system-level quality assurance for multi-step agentic workflows.

• ​Addressing the post-deployment human and behavioral factors that determine actual system adoption and efficacy.

​Attendees will gain technical insights into building compliant, deterministic, and highly resilient AI architectures ready for real-world operations in enteprise systems.

​​Agenda

​We begin at the macro-compliance level, establishing the legal, regulatory, and quality assurance frameworks required before production engineering starts. We then transition into micro-level technical architecture, focusing on the specific QA challenges of verifying complex, multi-agent system trajectories. Finally, we move to post-deployment operations, analyzing the behavioral factors and human-system interactions that dictate whether a technically sound system succeeds or fails after release.

​⬢ 18:00 – Doors Open

​⬢ 18:30 – Talk 1: From PoC to Production: A Practical Roadmap for Enterprise-grade AI Compliance in the Automotive Industry (Alireza "Andrew" Sharifikia)

​⬢ 19:00 – Talk 2: System Quality for Multi-Agent AI Systems (Raphael Aubel)

​⬢ 19:30 – Talk 3: Why AI Systems Fail After Deployment: The Missing Human Layer (Tomomi Tanaka)

​⬢ 20:00 – Networking Session

​⬢ 21:00 – Doors Close

​Speakers

​Talk 1 - From PoC to Production: A Practical Roadmap for Enterprise-grade AI Compliance in the Automotive Industry

​Speakers: Alireza "Andrew" Sharifikia (Software Engineering Manager at Corpy & Co.)

​Abstract: Today, nearly every major automotive supplier is experimenting with AI. However, moving from a successful Proof of Concept (PoC) to a mass-produced, safety-compliant product is a risky and costly challenge. About half of the companies that found the results of their AI investment to be underperforming cite AI governance as the leading cause; in this talk, we'll break down the components of AI governance, covering AI quality assurance and safety guidelines in the context of an enterprise client in the Japanese Automotive industry.

​Bio: Senior Manager in Software Engineering with a background in AI, Computer Vision, and Data Science. I started software development as a hobby, and today I’m a Senior Group Product Manager. I have worked for many great companies and teams over the past decade. I love contributing to the startup communities, and I am currently employed by Corpy & Co., where I lead the XAI Initiative and our collaboration with the TAI community.

​Talk 2 - System Quality for Multi-Agent AI Systems

​Speakers: Raphael Aubel (R&D Engineer at Corpy & Co.)

​Abstract: As the industry gradually moves from single-prompt AI applications toward more complex agentic systems, quality assurance must expand its scope accordingly.

Agents can call tools, update state, follow policies, and make decisions across multi-step workflows. As a result, correctness cannot be judged only from an isolated component test or a single response. It must also be evaluated across the system’s full trajectory: what the agent observed, what it decided, which tools it used, and how reliably it behaved across repeated runs.

In this talk, we will extend core principles from QA4AI’s System Quality axis to agentic AI systems. Through the example of a retail AI system scenario, we will explore how QA practices should account for workflow behavior, intermediate decisions, repeatability, and end-to-end reliability before such systems can be trusted in actual production workflows.

This session is intended for QA engineers, AI engineers, and technical teams interested in evaluating and building reliable agentic systems, as well as anyone curious about the potential practical issues emerging from agentic AI integration.

​Bio: Raphael Aubel is an R&D Engineer at Corpy&Co., where he works on applied AI projects from technical exploration to production-oriented delivery. His work has included AI/VLM pipelines for logistics document automation, QA4AI-based review of machine learning systems, vehicle-routing optimization for logistics operations, and computer vision pipelines for industrial inspection. Raphael focuses on practical reliability concerns in production AI, including evaluation, constraints, and alignment with real operational requirements.

​Talk 3 - Why AI Systems Fail After Deployment: The Missing Human Layer

​Speakers: Tomomi Tanaka (Founder, Behavioral AI Lab)

​Abstract: An enterprise AI system can achieve excellent benchmark performance and still fail completely after deployment. In many cases, the problem is not the model itself, but the human behaviors surrounding it: trust, incentives, cognitive overload, decision-making, and adaptation.

In this talk, I will explore the “missing human layer” through examples from enterprise AI adoption, where I have observed trust failures, incentive misalignment, and user adaptation patterns that technical evaluation and QA processes often fail to capture. Drawing from my experience across behavioral science, platform safety, and large-scale AI deployment, I will discuss why human behavior is often the biggest bottleneck, and opportunity, in AI adoption and impact.

The session will introduce a behavioral framework for designing AI systems that people actually use, trust, and benefit from at scale.

​Bio: Tomomi Tanaka is a behavioral scientist and AI strategist working at the intersection of AI systems, human behavior, and decision-making. She is the founder of Behavioral AI Lab, a Senior Behavioral Advisor at BEworks, and an adjunct professor at Kindai University.

Her experience spans large-scale platforms, including Amazon, Uber, and Match Group, where she founded Safety by Design to protect millions of users; international policy organizations, including the World Bank and the United Nations; and behavioral economics, with publications in the American Economic Review and Harvard Business Review.

She has given talks and workshops at UN Headquarters and other international forums on AI governance, digital safety, and behavioral design.

​​Organizers

​Ilya Kulyatin is an entrepreneur with work and academic experience in the US, Netherlands, Singapore, UK, and Japan. He holds a BA in Economics, an MA in Finance, and an MSc in Machine Learning. He's a 3x founder, now helping Japan grow the local AI ecosystem through a not-for-profit community, Tokyo AI (TAI), while building an AI-native system integrator and solutions provider, Foundry Labs株式会社.

​Alireza "Andrew" Sharifikia is a Senior Manager in Software Engineering with a background in AI, Computer Vision, and Data Science. I love contributing to the startup community, and currently lead the XAI initiative and AI product sales at Corpy & Co.

​​Supporters

​Tokyo AI (​​​TAI) is the biggest AI community in Japan, with 4,000+ members mainly based in Tokyo (engineers, researchers, investors, product managers, and corporate innovation managers).

​​​Privacy Policy

​We will process your email address for the purposes of event-related communications and ongoing newsletter communications. You may unsubscribe from the newsletter at any time. Further details on how we process personal data are available in our Privacy Policy.