Carevra is not a prompt with a resume pasted into it. Every document follows the same four steps, in the same order, for the same reason: the diagnosis has to come before the rewrite, and the facts have to come before the claim.
Your resume is read twice: once the way an applicant tracking system parses it, once the way a recruiter skims it in six seconds. We separate the two, because passing a parser and convincing a human are different problems.
Output: a scored read across keyword alignment, structure, evidence, positioning, seniority signal, and readability — with the specific lines responsible for each deduction.
Most resumes are underwritten, not badly written. Before rewriting, Carevra asks for the facts already true about your work — team size, budget, systems owned, results delivered — that your current draft leaves out.
We never invent a number. If you don't supply it, it doesn't appear. Everything else is a reframe of what you already wrote.
Your experience is rewritten toward one target role, in the register of your profession and the arc of your career — an operator reads differently from a clinician, and a Director reads differently from a senior IC.
Output: an ATS-safe resume, a matching cover letter, interview material drawn from your actual record, and — on the Executive Career Sprint — LinkedIn positioning and recruiter outreach.
The same resume against the same role produces the same score, every time. Scoring is deterministic and computed from measurable properties of your document — not a number the model is asked to guess.
You can re-run the analysis after any edit and see exactly which dimension moved, and by how much.