llama4aj brings local ChatGPT to Android

This is less a polished end-user product than a practical bootstrap kit for on-device AI apps, and that makes it genuinely useful to developers who want to avoid cloud APIs. It lowers the barrier from native LLM plumbing to a copyable Android/Java template, which is often the hardest part. Offline inference is the real selling point here: privacy, no per-token costs, and a path for apps that need to work without network access. The project still feels early-stage, with upstream syncing, broader platform testing, and documentation explicitly listed as TODOs. Because it sits on top of llama.cpp bindings, its long-term value depends as much on backend performance and device support as on the Java layer itself.