EDGE FDE
AdvancedTurn feedback into a better local model
Improve quality from real use without sending raw user text to a public trainer by default.
Signals to synthetic data to LoRA to a baked Core AI export, with honest privacy limits.
Collect Swift signals on device. Train and export in Python on a Mac. Load the bake in Swift. iOS 27 and macOS 27 only.
Lessons
Key concepts
- The loop is product signal, synthetic rows, LoRA on a Mac, Core AI
.aimodel. - Foundation Models is the system model API. Core AI ships adapted weights you bake.
- Default telemetry is metadata, not raw transcripts.
- There is no documented hot-swap LoRA. A new bake is an app release.
Takeaways
- You can collect Swift signals on device, train in Python on a Mac, and load the bake in Swift.
- You can improve quality from real use without sending raw user text to a public trainer by default.
- You ship a merged
.aimodelthrough Xcode 27, not a runtime adapter.
Agent briefs
Related courses
LLM LoRA for your writing style · Test AI behavior before you ship · Version prompts like product copy · PyTorch to Core AI in Xcode · Core AI vs Core ML vs MLX