Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn essential patterns for building reliable, production-ready LLM applications through this conference talk that challenges conventional agent frameworks and presents practical principles for creating truly effective AI agents. Discover why most "AI agents" in production are actually deterministic code with strategically placed LLM components rather than traditional prompt-and-tool-loop systems, and explore how successful founders are building impressive AI products by rolling their own stacks instead of relying on popular frameworks. Examine the gap between framework-based approaches like CrewAI, LangChain, and LangGraph versus what actually works in customer-facing production environments, and understand why the most effective agents are comprised mostly of traditional software engineering principles. Gain insights from real-world experience with various agent frameworks and learn the core principles needed to build LLM-powered software that meets production quality standards for end users.
