Skip to content

About the Generative AI Patterns Website

Just like object-oriented software needed Design Patterns, we now also need a robust collection of patterns to help enterprise architects leverage the power of Generative AI tools and knowledge graphs to have a large positive impact in the organizations they serve.

With the introduction of generative AI tools and agents, the art of building sustainable software is changing faster than ever before. To build applications that can scale out to meet the demands of large organizations, we need to understand the architectural trade-offs of different approaches quickly. We believe generative AI can help us quickly explore these options and create compelling interactive websites that help our stakeholders understand these patterns. These discussions need to be efficient. The best way to increase the bandwidth of these discussions is to create a shared understanding of common architectural approaches.

Referencing named patterns is one of the most essential tools to create shared understanding. Using named patterns started in 1977 with the publication of Christopher Alexander's groundbreaking book A Pattern Language: Towns, Buildings, Construction. Software architects adopted design patterns in the Gang-of-Four Design Pattern book in 1984. In 2003, the Enterprise Integration Patterns book also applied design patterns to computer integration.

Now, we are facing the most significant shift in application development for the last 20 years: the rise of generative AI. To stay productive, we need a pattern language for generative AI to allow our architects to discuss the relative merits of different approaches. This book attempts to codify many of the best practices in this rapidly changing field and give the reusable solution names that help increase the bandwidth of these discussions.

Creating a new pattern language is a challenging task. Generative AI applications are being built today with experts from different fields with distinct backgrounds. Our goal is to take a holistic systems-thinking approach to GenAI Patterns and include input from various groups, including:

  1. Application Architects
  2. Data Scientists
  3. Natural Language Processing Experts
  4. Rule Engine Architects
  5. Security Architects
  6. Ethics Review Boards

By sharing terminology, the architectural review processes will go smoother, resulting in better designs.

As of September 2025 we have identified the following top trends:

  1. Agentic AI as a core enterprise capability
  2. Continuous, real-time enterprise GenAI architecture
  3. AI-augmented EA tooling - custom diagram generators
  4. Enterprise-wide AI governance (security, legal, resp. use)
  5. Data fabric & knowledge graph backbones for agent traversal
  6. Open, modular AI reference architectures (layers etc.)
  7. Core modernization with an “intelligent core” knowledge graph
  8. Value-linked roadmaps & portfolio rationalization
  9. Pragmatism on impact metrics
  10. Skills & methods refresh within EA frameworks

Before we introduce you to the GenAI Patterns, we need to make sure everyone has a strong understanding of the key concepts that underpin Gen AI solutions. We encourage you to first visit the GenAI Patterns Concepts section of the website.