Modern Microservices for Enterprise AI: Patterns, Pitfalls, and Practices

AI is quickly becoming a must-have for businesses that want to stay competitive. To reduce time-to-value, most efforts involve adding AI features to existing apps using existing frameworks. Microservices can be a great approach for this. Structuring your app as microservices make it easy to incorporate new functionality, mix languages (like Java for the core business logic and Python for AI), and makes it simpler to update new AI features as needs change.

In this session, we'll share what we've learned working with developers on real enterprise projects powered by AI. We’ll talk about practical topics like figuring out the right number of microservices, keeping up with fast-changing AI models, designing event-driven systems so agents can talk to each other, and managing workflows that blend traditional code with AI-driven components. We’ll also touch on smart ways to handle your data, so you get the right balance between isolation and sharing - while making sure everything stays consistent. Along the way, we will cover design practices that make it easier to build with AI. This include organizing services so you can trust AI-generated code and tips for minimizing hallucinations.

While we won’t have live demos during the session, we’ll share links to hands-on labs and code examples you can try afterwards. Our examples will use Spring, Java, and Oracle Database, but these ideas work across lots of tech stacks and programming languages.