Generative AI enables powerful new capabilities, but they come with some serious limitations that need to be addressed to deliver a reliable application or agent.

Experts in the field have compiled a library of 32 proven design patterns to solve the challenges typically encountered when developing applications with LLMs, such as hallucinations, nondeterministic responses, and knowledge cutoffs.

This book consolidates research and real-world experience into practical advice that can be integrated into projects. Each pattern describes a problem, demonstrates a proven solution with a fully coded example, and discusses trade-offs.

Key topics include:

• Designing around the limitations of LLMs
• Ensuring that generated content adheres to a specific style, tone, or format
• Maximizing creativity while managing different types of risks
• Building agents that plan, self-correct, take action, and collaborate with other agents
• Combining patterns into agentic applications for various use cases

Pages: 496