How This Former Poker Pro Runs an AI Company With Just 1 Meeting Per Week

**George Kurdin** spent years as a professional poker player before online poker was shut down overnight. The skill he carried into building a company wasn't the one you'd expect. It had nothing to do with reading faces across a table and everything to do with expected value, risk, and the discipline to iterate before you get crushed. Now George is co-founder and CEO of Monk, an AI-native accounts receivable platform automating the whole contract-to-cash lifecycle. He sat down with Brennan Pothetes to make the case for building in the least glamorous corner of AI, why "touching money" earns you the right to build something big, and why he's genuinely worried about the foundation model labs even as he builds in their blast radius. This one gets into the real mechanics of building an AI company in 2026: hiring 1 in 700, running on one meeting a week, what an "agent harness" actually is, and where defensibility still lives once the models are a commodity. **Inside the episode:** - Why George says poker teaches risk and EV, not reading people, and how that maps to founding <!--THE END--> - The principles he used to pick his market, including "don't get nuked by Satya and Sam" <!--THE END--> - Why he believes you "earn the right to build a big business if you touch money" <!--THE END--> - His spicy take on the labs, valuations, and the shift toward Chinese and open models <!--THE END--> - What an "agent harness" is, explained through L4 self-driving cars and human-in-the-loop <!--THE END--> - How he hires (1 in 700) and runs the team with one weekly all-hands and no standups <!--THE END--> - Where moats still exist in the AI app layer when the tech is no longer the advantage Subscribe for more conversations with the founders building what's next. Watch on YouTube, listen on Spotify and Apple. Follow George and Monk for more on AI-native finance.