AI Agents and Prediction Markets
AI Agents and Prediction Markets
How can AI agents help us make better forecasts, and what does it take to build a prediction-market product across the US and Japan?
Join Tokyo AI (TAI) for three short talks and a 45-minute panel. We will look at the US regulatory questions facing prediction markets, how AI agents participate in these markets, and how two Japanese founders build prediction products.
For AI builders, founders, researchers and product or compliance teams interested in prediction markets. No prior knowledge of the field is required.
When: Tuesday, October 13, 2026, 18:00-21:00 JST. The evening time window is provisional.
Where: Tokyo, venue to be announced.
Agenda
Provisional clock times. The three talks are 15 minutes each, followed by a 45-minute panel including audience Q&A.
18:00 - Doors open
18:30-18:40 - Welcome and introduction
18:40-18:55 - Talk 1: US Prediction Markets: The Year Ahead (Dan Ullman)
18:55-19:10 - Talk 2: What Happens When AI Trades the Future? (Takanori Shirasaka / shora, POYP)
19:10-19:25 - Talk 3: MIRAIMA: points, participation and collective forecasts (Keita Adachi; topic pending)
19:25-20:10 - Moderated panel and audience Q&A
20:10-21:00 - Networking
21:00 - Doors close
Talks
Talk 1: US Prediction Markets: The Year Ahead
Proposed speaker: Dan Ullman, partner at Orrick.
Bio: Dan advises crypto, fintech and gaming clients on regulatory and enforcement matters. He spent 16 years at the US Commodity Futures Trading Commission, most recently as an Associate Director in its Division of Enforcement. His work covers prediction markets and event contracts.
Description: Prediction markets have taken the US by storm, capturing the imagination of market participants while raising questions from regulators and state authorities. Dan will explore where the US currently stands with respect to Prediction Markets, what’s currently happening in this space from a market and regulatory perspective, and what we can expect to evolve over the next 12 months.
Talk 2: What Happens When AI Trades the Future?
Speaker: Takanori Shirasaka / shora, founder of POYP.
Bio: Takanori is the founder and CEO of POYP, a prediction-market app in Japan built around points rather than money. He previously served as CTO of an AI startup, where he led machine learning and computer vision products, and studied machine learning at the University of Tokyo. His work focuses on market design, prediction markets as information infrastructure, and the legal constraints on running them in Japan.
Description: AI agents now account for a large share of trading on major prediction markets, yet frontier models given real capital in 2026 lost money across the board. Takanori will examine what actually changed when AI started trading the future: who participates, who wins, and why forecasting accuracy does not translate into returns. He will then explain how POYP, a prediction market built in Japan, treats AI traders, and what that choice reveals about who a market is for.
Talk 3: MIRAIMA: points, participation and collective forecasts
Speaker: Keita Adachi, founder of MIRAIMA.
Description: How can a points-based prediction product invite participation in Japan, and what can an operator learn from its collective forecasts? The final title and abstract will be set with Keita.
Panel
When does AI improve a crowd forecast, and when does the crowd know something the model misses? The panel will compare product choices, market integrity, forecasting quality and the different questions builders face in the US and Japan. The audience Q&A is part of the 45-minute panel. Additional panelists will be announced only after their participation is confirmed.
Organizer
Tokyo AI (TAI) connects engineers, researchers, founders and other people building AI in Japan through talks and community events.
Privacy Policy
We use registration details to manage this event and communicate with attendees. If you opt in to future updates, you can unsubscribe at any time. Read TAI's Privacy Policy.