In a Stack Overflow blog post, Ryan interviews Leo de Moura about using the Lean theorem prover to formally verify AI agents, ensuring correctness beyond probabilistic reasoning. The discussion highlights how automated reasoning complements probabilistic AI models and enables continuous code optimization through AI