Version 1.2.4 - Matchflo now checks whether on-device AI can run before you start an analysis. If it can't, you'll see a notice under the Analyze button explaining why, instead of an error partway through. - Clearer explanations when your Mac or device doesn't support Apple Intelligence, including which hardware is required. - On Apple silicon Macs where Apple Intelligence is turned off or not available in your region, Pro users can analyze with the downloadable Advanced AI model. - Clearer messages when Apple Intelligence's on-device model is still getting ready, so you know what's happening instead of seeing a generic error.

• Clearer messages when Apple Intelligence's on-device model needs a moment to finish getting ready, so you know exactly what's happening instead of seeing a generic error.

Reliability - Analysis and cover letter generation now fail fast with a clear message if a prompt is too large, instead of hanging or throwing a generic error after a doomed request. - Both features now time out after 45s instead of potentially hanging forever if Apple Intelligence stalls. - Added recovery logic so a truncated/malformed model response gets a second repair attempt before giving up. Correctness - Cover letters could come back with stray artifacts the model added despite instructions — a leading "Here's your cover letter:", wrapping quotes, or markdown code fences (including the literal word "markdown" leaking into the text if the model used a ```markdown fence). All of that is now stripped from every generation path, not just section regeneration. Visibility - Added structured logging around every prompt sent / response received, so failures are debuggable via Console logs without exposing resume/job description content in production builds (full content only logs in DEBUG builds).