Building mobile applications that consume predominantly non-deterministic data pose a new and unique set of challenges. This talk will review tactics for handling non-deterministic data in order to build resilient and reliable mobile front ends. Key takeaways will include: how to optimize use of Dart Models for validation, defaults and sanitation to maintain a solid contract for data shape, type and size; how to implement good prompting and OpenAI function calls to get more deterministic data; and how to create more flexible UI widgets that can handle a broader range of content sizes. This talk will finish by taking a high level view of how the widespread availability of non-deterministic data from APIs like OpenAI will change how we develop front ends.