TEQnation 2024: Lessons learned from building a Generative AI playground – François Bouteruche

Generative AI, there are those who talk about it and those who build with it. In this session, I share the lessons learned from building the [.NET FM Playground](https://github.com/build-on-aws/dotnet-fm-playground). This project has started as a way to educate myself on this broad and new domain. No better way for me to learn than getting my hands dirty. During the session, you get a hands-on, developer oriented understanding on how to use Generative AI in your application. You won’t hear about training large language models (LLMs), I’m no ML expert. We walk through the different playgrounds to understand a few key concepts. The Text playground helps us to make our first steps with LLMs and do a bit of prompt engineering. The Chat playground surfaces why and how you can memorize your conversation. The Voice Chat playground showcases how you can mix and match different AI technologies to deliver voice-based interactions. The Agent playground illustrates how you can use Retrieval Augmented Generation pattern to overcome LLMs knowledge limitation and how you can make them accomplish complex tasks. When you walk out from this talk, I hope you go build your first Generative AI enabled app.