Agentic Economy Benchmarking MAS
As autonomous agents become more capable and prevalent, a key question emerges: how do we measure and trust their performance in complex, human-relevant environments?
This closed-door workshop brings together researchers, industry leaders, and policy thinkers to co-design a verifiable sandbox — a simulated trading environment that tests multi-agent systems under realistic yet controlled market conditions.
Participants will explore how an “esports-style” benchmark can make agentic systems auditable, reproducible, and comparable across technical, economic, and social dimensions.
The goal is to collaboratively shape Whitepaper v0.1: Verifiable Sandboxes for Benchmarking Multi-Agent Systems, defining shared evaluation frameworks for performance, adaptivity, cooperation–competition, fairness, transparency, and safety.
Timeline:
13:00 – 13:30
Registration & Networking
13:30 – 13:45
Welcome Remark
Yiju Jia, Tensorlink
13:45 – 14:30
Keynote – Multi-Agent System Design and Human–AI Collaboration
Prof. Per Ola Kristensson, University of Cambridge
14:30 – 15:00
Fireside Dialogue I – From Academic Testbeds to Industry Platforms Prof. Philip Torr (Oxford) × Frank O. Miller (Colt, Chief AI Officer)
15:00 – 15:30
Tea Break
15:30 – 16:00
Fireside Dialogue II – Agentic Trading and the Sandbox Economy
Christopher Hayes, Founder of Guerrilla Quant · Dequn Teng - University of Cambridge Moderator: Yiju Jia
16:00 – 16:30
Academic Talk II – Economic Mechanisms and Agentic Design
Dr. Michel Ferreira Cardia Haddad, Queen Mary University of London / Cambridge Affiliate
16:30 – 17:00
Fireside Dialogue III – Trust, Safety & Governance in the Agentic Economy
John Walker-Robertson (LSEG) · Lily McCann (WPP Media) · Xiangjian Jiang University of Cambridge
Moderator: Yiju Jia
17:00 – 17:30
Talk – Strategically Innovating with Agentic AI: Moving Beyond Short-Term Tactical Wins
Dr. Gary Fox, CEO, Business Ecosystem Design Labs
17:30 – 18:00
Talk – AI Safety and Governance for the Agentic Era
Dr. Matthew Barker, Cambridge University / Trustwise
18:00 – 18:30
Closing Talk – Somebody Needs to Receive That Pain: A Position on Agentic Accountability Amber Hu University of Oxford
18:30 – 19:00
Drinks & Mingle
Why attend?
Engage directly with leading voices from academia and industry.
Help define the next generation of multi-agent benchmarking standards.
Contribute to a cross-sector dialogue on trust, safety, and governance in the emerging Agentic Economy.
Be part of a research-building community shaping how autonomous systems interact, compete, and cooperate in the real world.