Coursera Flash Sale
40% Off Coursera Plus for 3 Months!
Grab it
Learn twelve essential engineering principles for building reliable, production-ready LLM-powered applications in this 17-minute conference talk. Discover why most successful AI products use deterministic code with strategically placed LLM components rather than following traditional "prompt plus tools" agent patterns. Explore practical factors including natural language to tool call conversion, prompt ownership, context window management, structured output handling, state unification, execution control through simple APIs, human-in-the-loop integration, custom control flow design, error compaction strategies, focused agent architecture, flexible triggering mechanisms, and stateless reducer patterns. Gain insights from extensive experience testing various agent frameworks and conversations with founders building customer-facing AI applications, with accompanying resources including code examples and community discussions.
