This talk introduces a crucial but often overlooked aspect of AI application development: evaluation-driven development (EDD). Using a simple LangGraph agent as a practical example, we'll demonstrate why and how to build a robust evaluation framework that goes beyond simple unit tests. We'll explore the importance of continuous evaluation during the development cycle and how this practice directly translates to the need for comprehensive observability in production, ensuring your AI application remains accurate, reliable, and effective in the real world.

AI agents are everywhere, but how do you go from demo to dependable? In this talk, we’ll build agents using LangGraph, and take it beyond toy examples into enterprise-grade readiness.

We’ll start with context engineering concepts like: control flow, prompt engineering, prompt management and tool calling. From there, we’ll build an agent step-by-step, enabling contextual reasoning via Retrieval-Augmented Generation (RAG), and integrating tool use for dynamic task execution. We'll also cover agent observability with open source observability tools such as LangFuse to bring full observability into the decision our agents make.

Whether you're exploring AI agents or planning your first production deployment, you'll leave this session with the patterns, tools, and clarity to do it right.

How can we expect LLM products to impact software engineering?

Join Chelsea Troy for an examination of how LLM products are built. She'll walk through the domain: what data LLMs are trained on and how they're trained, and spell out how those factors impact what LLMs tend to do well with and where they tend to struggle.

Then, we'll talk about how that affects the design of the products built around LLMs, like ChatGPT, Gemini, Claude, Copilot, or Devon.

Finally, take a tour with Chelsea through current research elucidating the impact of LLM product use on developer workflows, and dig into what that means for us as practitioners in an LLM-enabled world.

Performant Architecture for a Fast Gen AI User Experience

More and more developers are being asked to build Generative AI experiences into their web applications everywhere. However, the user's experience with Generative AI can sometimes be slow and frustrating. It’s been years since we would gladly wait over 10 seconds for a response to anything online. As AI processing gets faster, there are new limiting factors that could be negatively affecting your AI user experience.

In this talk, we will cover different ways to minimise laggy experiences in web applications when interacting with third-party Generative AI tools. We’ll be using a practical demonstration to show how you can reduce latency as we build one of my favourite sci fi gadgets, the babel fish.