🎥 Recorded live at the MLOps World | GenAI Summit 2025 — Austin, TX (October 8, 2025) Session Title: Agents as Ordinary Software: Principled Engineering for Scale Speaker: Linus Lee, Entrepreneur in Residence & Advisor, AI, Thrive Capital Talk Track: Agents in Production Abstract: How do you scale an AI agent that autonomously handles thousands of research and automation tasks weekly—without breaking? In this session, Linus Lee shares how Thrive Capital’s in-house research engine, Puck, executes thousands of tasks—from surfacing current events to drafting reports—through a robust orchestration library called Polymer. With fewer than 10 engineers, the team sustains this scale by leaning on four core engineering principles: composability, observability, statelessness, and changeability. Linus offers a behind-the-scenes look at the system’s architecture, walking through design patterns and practical implementation examples that make complex agentic systems maintainable, testable, and scalable. Attendees will see how classical engineering discipline can co-exist with frontier LLM innovation. What you’ll learn: • Design patterns for building robust, composable LLM systems at scale • How to apply principles like composition, adapters, and stateless effects in production • The importance of observability and traceability in multi-agent orchestration • How to evolve systems seamlessly as new model capabilities emerge • Why classic software engineering values matter more than ever in the AI era.