The One About Agents & Governance
What happens when AI agents interact at scale, and how do we safely deploy them in production environments? Explore the fascinating intersection of agent communities and enterprise deployment, examining both how agents behave in large-scale ecosystems and the critical governance frameworks needed to harden them for real-world business applications.
More About the Sharings
Dr. Wenxuan Zhang (Assistant Professor, SUTD) will share on "MoltNet: Understanding Large-Scale AI Agent Communities"
How do AI agents behave when they interact with each other at scale? Dr. Zhang will present MoltNet, an empirical study examining whether agent-agent interactions on the MoltBook platform reproduce core mechanisms observed in human social systems. Using early-2026 platform data and grounding the analysis in sociological theory, he'll explore agent behavior across four key dimensions and share insights for designing and governing large-scale agent ecosystems. He'll also highlight complementary work including Auto-Arena, a fully automated LLM evaluation framework, and DR-Arena, a dynamic platform for assessing research agents through evolving real-time tasks. (Technical Level: 200)
Michael Hart (Founder, Denvelop Private Limited) will share on "Hardening Agentic Solutions for Production Deployment"
Deploying agentic solutions like OpenClaw in production involves far more than secure containerisation. Michael will cover critical considerations across infrastructure, authority, and influence, demonstrating how these factors apply across different business use cases. Drawing from his experience building Tropic, a security control plane for OpenClaw, he'll share practical approaches to production-ready agent deployment and the governance challenges that emerge when agents move from development to real-world business environments. (Technical Level: 200)
Frederick Liau (Founder, WorldMind) will share on "The AI Governance Gap: When AI Outputs Become Institutional Decisions"
When AI-generated errors appear in court filings, the real challenge isn't AI hallucination—it's how those errors travel through organizational decision chains to shape institutional outcomes. Drawing from recent cases, including Singapore lawyers fined for submitting fictitious AI-generated legal citations, Frederick will introduce the concept of Systemic Decision Rot—how AI-generated insights distort through multiple layers of human interpretation. He'll demonstrate how small AI inaccuracies compound through decision pipelines and why model safety frameworks are insufficient once AI influences institutional decisions. The talk concludes with DecisionAI, an architectural layer providing verification, policy enforcement, and traceable decision logic to protect decision integrity. (Technical Level: 200)
More About the Speakers
Dr. Wenxuan Zhang is a tenure-track Assistant Professor at Singapore University of Technology and Design (SUTD). He received his PhD from the Chinese University of Hong Kong and previously worked as a research scientist at Alibaba Singapore, earning the Ali Star award. His research focuses on natural language processing and large language models, with emphasis on inclusive, multilingual models and trustworthy AI systems. He has co-led influential open-source projects including SeaLLMs, Babel, and AutoArena, and regularly serves as an area chair for leading AI conferences including ACL, EMNLP, NeurIPS, and ICLR. He is recognised among the World's Top 2% Scientists by Stanford and Elsevier in 2025.
Michael Hart has spent 17 years in the technology industry, including 6 years as CTO and CPO at Sleek, where he led product and technology through the company's Series A and B funding rounds. He now runs Denvelop Private Limited, building AI solutions and advising SMEs on AI and technology initiatives. Michael is currently developing Tropic, a security control plane for OpenClaw, and serves as VP of Partnerships at Click Global Talent, helping SMEs worldwide scale through global talent networks.
Frederick Liau is the founder of WorldMind and the creator of the DecisionAI framework, which focuses on the architectural layer required to ensure trustworthy institutional decisions in the age of AI. His work explores the intersection of AI systems, organizational decision pipelines, and governance structures needed to safely deploy probabilistic AI in real-world institutions. Frederick is also the inventor of a patented system underlying the DecisionAI architecture.
More About the Series
AI Wednesdays is Lorong AI’s weekly gathering, bringing together practitioners, researchers and innovators for technical discussions on research insights, product development and engineering practices.
Get involved: Learn more about Lorong AI | Speaker Sign-up | WhatsApp Community | LinkedIn | X