AI Disclaimer

1. What the AI is doing

• Analyses — comparing your CV text against the job description text, returning a Match Score, an ATS Probability, and a list of keyword / formatting gaps.
• Auto-revisions — rewriting your bullets to inject missing keywords and tighten phrasing. Every injected phrase is tagged [REVIEW: …] so you can audit and accept or reject each change.
• Blueprint validation — after a revision, a second AI pass checks whether the items the analysis flagged were actually addressed in the new CV.

2. What the AI is NOT doing

• It is not predicting whether you will get an interview, a callback, or an offer. The Match Score and ATS Probability are calibrated estimates, not the actual scoring used by any third-party recruiting software.
• It is not inventingqualifications. If the AI adds a phrase referring to experience you don't have, that is your responsibility to remove before sending the CV out. The [REVIEW:] tags exist precisely to make this obvious.
• It is not a substitute for your judgment. The AI doesn't know your career goals, your visa situation, or whether the role is actually right for you.

3. Hallucinations and how we limit them

Large language models can occasionally produce text that sounds plausible but is not grounded in the input. We take three practical steps to limit this risk:

• We flag every AI insertion with the [REVIEW:] tag inside the revised CV. You see exactly what the AI added, in context, before exporting.
• We never feed your private notes into the model. Notes you attach to a scan are stored alongside the scan but are never part of any AI prompt.
• We validate the revision against the original analysis blueprint, so you can see at a glance whether the AI actually addressed each gap or merely shuffled words.

4. Your responsibility before sending the CV

Treat the revised CV as a draft, not a finished product. Before you send it to a recruiter or upload it to a job board:

1. Read every [REVIEW:] tag and decide whether the claim is something you would say in an interview.
2. Replace any sentence you can't back up with concrete experience. Removing an AI suggestion is always cheaper than being caught out in an interview.
3. Run a final spelling / formatting check in your own editor — our exporter is good but not perfect, especially on heavily customised templates.

5. Bias and fairness

AI models can carry the biases of the text they were trained on. We avoid embedding demographic signals (gender, ethnicity, location nuances) in our prompts, and we score on keyword-relevance and seniority signals only. If you ever feel a score doesn't reflect reality, please contact us — we log unusual scoring patterns so we can audit and fix them.

6. Where this disclaimer fits

This page complements our Terms of Service (the legal contract) and our Privacy Policy(what data we process). The Terms control if there's ever a conflict; this page is the human-friendly companion.