Confirm access
Enter the email BetterWork has on file for this candidate.
Enter a well-formed email address.
✓ Looks good. Press Continue.
Inputs
Choose the résumé you’d like the tool to work with, drop the file on the box below, or copy and paste the text from the file.
Upload Word (.docx), PDF or plain text — choose a file or drop it on the box below. For Apple Pages, export to Word or PDF first. Scanned PDFs can't be read; paste the text instead.
If you have career information beyond what’s in your résumé, add it here. This could come from a career log, brag sheet, or performance review.
Upload Word (.docx), PDF or plain text — choose a file or drop it on the box below. For Apple Pages, export to Word or PDF first. Scanned PDFs can't be read; paste the text instead.
Copy and paste the text from your target job description. Do not include links.
Your inputs changed, so the previous analysis was cleared. Generating will analyse again.
Gap analysis
What the role needs, where your materials show it, and where the gaps are. Gaps never block generation.
Review warnings
Revise
Say what to emphasize or reword, or correct a fact. A correction is saved to your additional context and takes precedence over your résumé. Employers, titles and dates always come from your résumé.
Download
Debug trace
Staff only. Every run and every model call this session: stages, timings, sizes, error codes, and the full prompt and reply. The trace holds the candidate's material, so share it only with the BetterWork team.
Prototype diagnostics — estimated cost this session: ~$0.000 (0 calls)
Estimated from character counts. The artifact runtime does not report token usage. The viewer's own Claude account pays for these calls. This panel is for the prototype walkthrough and will not appear in the delivered tool, which records real token counts server-side.
| Call | Tier | In (chars) | In (~tok) | Out (chars) | Out (~tok) | Time | First text | Est. cost |
|---|
Analysis payload breakdown
Where the most recent successful analysis call's output actually went, measured from the parsed response object — not the prompt, the result.
| Part | Chars | % of total | ~Tokens | ~Cost |
|---|
The platform chooses the model for each tier and may substitute a cheaper one — these rates are editable so they can be corrected during the walkthrough (USD per 1,000,000 tokens). Not saved.