Seeking Balance The Future of Poker Strategy (Start Learning this Now)
Feb 13, 2026
Philippe Beardsell, co-founder of Ruse AI and lead researcher at GTO Wizard, builds poker-solving AI. He discusses advances in multiway solving and QRE regularization. He talks about why ICM is uniquely tricky, how player profiles and exploiter tools can change strategy, and scaling solvers to PLO and large games.
AI Snips
Chapters
Transcript
Episode notes
From Startup Dream To GTO Wizard Researcher
- Philippe recounts his original dream to build superhuman poker AI and how joining GTO Wizard gave more resources to pursue it.
- He moved from a two-person startup to a larger team, enabling deeper ML and product focus.
Explore Rare Actions With Low Frequency
- Embrace small, regular exploration when studying or playing to find unintuitive but good plays.
- Philippe and Jared recommend trying weird actions rarely (e.g., 1%) to discover viable sizes or lines without major win-rate risk.
Multiway Solving Is The Next Big Foundation
- Philippe: multiway (3+ players) is the main development focus; three-way was a proof of concept with plans to scale to 9 players.
- He views multiway as foundational for profiles, human modeling, and other features.