CoverSentry

CoverSentry targets a specific modern problem: applicants increasingly draft cover letters with a chatbot, recruiters increasingly screen for it, and neither side has a reliable way to tell how a specific letter reads. The tool takes an uploaded or pasted letter and returns a confidence score with a three-way verdict — human-written, mixed or AI-generated — plus recommendations for fixing the passages that flag highest.

The engine is a fine-tuned BERT language model trained on a custom dataset of cover letters rather than general web text, which matters because cover letters have their own conventions and clichés. Processing happens in memory on the servers and documents are discarded immediately after analysis, and the site states that no tracking cookies are used and no data is sold. A companion free CV builder generates ATS-friendly resumes with pre-written bullet points covering more than 1,200 job titles, downloadable as PDF or DOCX without watermarks or registration walls.

The project is unusually candid about the limits of AI detection. In an October 2026 update it reported that frontier models now write convincingly enough that their letters will most likely pass as human, that its detector can no longer make accurate predictions on that class of output, and that retraining would raise false positives. That transparency matters when you are deciding how much weight to put on any single detection score — treat CoverSentry as one signal among several, not an oracle.