It feels like the perfect time to launch this, so here we go 🗣️
After 15 years as an Android developer, including working at companies like X and Disney, I think I've developed a pretty clear picture of what Android teams and large codebases actually need day to day.
On top of that, I've spent a lot of time studying, experimenting with, and refining how we use AI coding agents for Android development over the past months.
The fact is that giving every developer an AI agent doesn't automatically make a team more productive. In fact, I've been seeing the opposite way too often:
- Devs burning huge amounts of tokens by repeatedly feeding agents the same context.
- Devs going through iteration after iteration instead of making agent output more deterministic from the start.
- Devs generating slop they don't properly review, curate, or sometimes even understand.
- Teams using generic prompts and agents for everything instead of specialized Android skills for specific tasks or the right tools.
- AI-generated tests that test implementation details and break after the smallest refactor.
- Devs generating very inefficient UI code always from scratch, and almost never reusing the existing components, neither following the project design system.
- Devs generating huge AI-powered PRs, and lots of them in a very short time, creating a massive review bottleneck that ends up hurting the whole team’s productivity.
And many more things.
There is definitely a much better way to work with AI agents.
That's why I'm launching Proficient Android AI 💯
I'm putting together what I've learned over 15 years of Android development, combined with everything I've learned about directing AI agents effectively, into a practical training program for individual Android engineers and teams.
The goal is simple and very straightforward: less waste, better code, more productive teams, and more proficient Android engineers ❤️🔥