Generative AI Design Patterns

Generative AI enables powerful new capabilities, but they come with some serious limitations that you'll have to tackle to ship a reliable application or agent. Luckily, experts in the field are already hard at work compiling a library of tried-and-true design patterns to address the challenges you're likely to encounter when building applications using LLMs and other GenAI models--hallucinations, nondeterministic answers, and knowledge cutoffs among them. You'll find 31 of the most essential here.

Authors Valliappa Lakshmanan and Hannes Hapke codify advances in research and real-world experience into advice that you can readily incorporate into your projects. Each detailed explanation includes a description of the problem, a proven pattern to solve it, an example, and a discussion of potential trade-offs. Whether you read it cover to cover for inspiration or use it as a daily reference, this practical guide will help you troubleshoot whatever problems may arise.

• Design around the limitations of LLMs, such as hallucination and nondeterminism
• Force LLMs to generate text that follows a specific style or grammar
• Maximize creativity while balancing different types of risk
• Extend the capability of an LLM beyond just content creation
• Use patterns together to solve a variety of different use cases