🎥 Recorded live at the MLOps World | GenAI Summit 2025 — Austin, TX (October 9, 2025) Session Title: How Math-Driven Thinking Builds Smarter Agentic Systems Speaker: Claire Longo, Lead AI Researcher, Comet Talk Track: Agents in Production Abstract: While everyone’s talking about LLMs, few discuss the mathematical principles that should drive how we design and deploy them. In this technical deep dive, Claire Longo, Lead AI Researcher at Comet, breaks down how math-driven thinking helps build smarter, safer, and more reliable agentic systems. Claire demystifies the mathematical foundations that underpin model architectures — explaining how understanding these structures helps us choose the right models, anticipate weaknesses, and design modular, production-ready agents. She clarifies what “reasoning” actually means in practice and introduces a rigorous benchmarking mindset based on statistical hypothesis testing. Using a real-world case study — a poker coaching agent that blends LLM reasoning with statistical performance models — Claire demonstrates how hybrid systems can combine reasoning models with mathematical algorithms, allowing each to do what it does best. What you’ll learn: • Why math is the missing language in today’s agentic AI design • How to use mathematical intuition to guide model selection and architecture decisions • How to anticipate weaknesses in LLM-based agents before they reach production • The principles of hybrid agent design and modular reasoning • A statistically sound approach to benchmarking and model comparison.