Outsmarting the genie: How to make smart architecture decisions when building Gen AI apps
The genie is out of the bottle. Generative AI is everywhere, and promising to fulfil your every wish. Just like wishes however, building generative AI apps can be risky if not well thought out. This session explores the architectural decisions and trade-offs you need to consider to build reliable and performant generative AI applications. We’ll look at what generative AI is (and isn’t), what factors drive model and tooling choices, techniques for enhancing safety and accuracy, considerations for testing and optimising, as well as responsible AI principles.
We’ll look at some real-life use-cases to help understand how to apply the best practices, and explore how to identify good and bad use-cases for generative AI. Whether you are playing about with POCs, or ready to jump into large-scale production apps, this talk will equip you with the architectural knowledge to "outsmart the genie", use those wishes wisely, and unlock the full potential of generative AI.
