This session talks about how we managed to reduce our app start up while increasing control and complexity of what we do when users open the app. The talks navigates around topic of caching strategy, configurations, managing concurrency, benchmarking and analysing data to take important decisions.

AI coding tools look impressive when they start with a blank project. But most developers do not work with blank projects. We work with years-old codebases, undocumented decisions, mixed architectures, custom conventions, technical debt and systems where one seemingly simple change can affect ten other things. What are specifics that a mobile applications specialist should know? So how do we actually make AI useful in this environment? This roundtable brings Android developers together to discuss how AI is changing everyday software development — from developers using AI for the first time to teams trying to integrate AI deeply into mature codebases. We will explore questions such as: •⁠ ⁠How does AI behave differently in a fresh project vs a large legacy codebase? •⁠ ⁠How much context does an AI agent need to understand your project? •⁠ ⁠What does good context engineering look like? How would it change for a mobile apps landscape? •⁠ ⁠Should teams maintain rules, architecture documents, examples, prompts, or agent instructions? •⁠ ⁠How do we evaluate whether AI-generated changes are actually correct? •⁠ ⁠How do developers avoid blindly accepting plausible-looking •⁠ ⁠How to keep up with rapidly changing Android and iOS apps and best practices for an AI enabled team? •⁠ ⁠What workflows are teams already using successfully? We will moderate the conversation, introduce discussion prompts, share some of our own experiences, and then bring the room into the discussion.