Today’s AI-assisted development works best for smaller and specific code. Microservices are a good way to structure an application into components that are more amenable to AI-assisted development. Furthermore, many AI projects involve integrating AI capabilities into existing apps, and microservices align well with this need.

In this session, we will share insights learned from working with developers building AI-powered enterprise apps. We'll cover topics such as determining the optimal number of microservices, integrating rapidly evolving LLMs and AI features, designing efficient event-driven architectures, managing transactions and workflows, and leveraging databases to balance isolation with data sharing while ensuring strong consistency.

Throughout this talk, we will discuss how to attack these topics with an eye to simplifying AI-assisted development. For example, by adopting patterns that make it easier for AI to generate code. The key concepts will be demonstrated using Spring, Java, and Oracle Database, though the principles apply broadly across other frameworks, languages, and databases.