Privacy

This page describes what CandidateFirst collects and how it is used, in plain language. If you only use part of the product, only the matching parts below apply to you.

Your account

• Your email address and a sign-in session, so you can log in.
• Usage events: which feature you used, when, which Claude model handled it (recorded as “none” when no model call was involved, as with a pricing-page visit), the input and output token counts, our estimated dollar cost, and, on every pricing-page visit while you're signed in, whatever value that page's ref URL parameter carried, or the word “direct” when there wasn't one, capped at 64 characters, so we can meter spend and see what drove the visit. A pricing-page visit also logs a second, placeholder feature name internally; it is not a record of you using that feature. Outside the URL value, these never contain the content of your materials.

Qualifying listings

• The job listing URLs you submit, and the qualification results, tied to your account.
• Other members see counts only (for example, how many members qualified a listing). Never your name, never which listings you looked at.

The profile you build (only if you use these features)

• Your resume file, and the work history, education, and skills we parse out of it, merged with the same fields from a LinkedIn export where you provide one.
• Writing samples you add and the tone profile derived from them.
• Your job preferences, stories, and experiences you enter.
• LinkedIn connections you import: name, role, company, profile URL, email, and the date you connected, where your export includes them. Used to show you your own connections at a company and to let you draft outreach to them. Never shown to any other member.
• Recommendations written about you by others on LinkedIn: the recommender's name, title, and what they wrote, imported as a writing sample where your export includes them. This is someone else's writing, not yours.
• Emails you choose to forward to your personal clipper address, parsed to update your tracker.
• Application materials the product generates for you (resumes, cover letters, outreach), saved to your library.

AI processing

To do the work you ask for, we send the text that feature needs to Anthropic’s Claude API. That includes the job listing you submitted, a summary of your profile (recent roles, skills, salary floor, and preferences) when we qualify a listing against it, the company name we are researching for you, your resume when you ask us to parse it, every writing sample you’ve saved (sent together, and re-sent each time you add a new one, to regenerate your tone profile), a PDF writing-sample file the moment you upload it during setup, to extract its text before you save anything (DOCX and TXT files are read locally, not sent to Anthropic), draft writing samples during setup once you’ve marked two as representative, to preview your tone before you save them, your full stored experience, education, and skills when you generate a resume, when you generate a cover letter, your four most recent experience entries (up to two bullets each), your stories, hiring-manager note, what you’re looking for next, your tone profile, and up to five of your recent cover-letter openings (no education or skills for this one), when you generate answers to application questions, those questions plus your full experience, stories, what you’re looking for next, skills, and tone notes, the emails you forward to your clipper address or paste in manually, when our regex parser can’t read them with confidence, a saved resume, cover letter, or essay answer when you save it, so we can classify its sections, a saved resume or cover letter specifically when you adapt a package for a new role, so we can rewrite its sections (essays are always regenerated fresh during adaptation, not reused), and the materials we draft for you, including outreach drafts, who they are addressed to, and, for outreach specifically, your current or most recent job title and employer plus one strength from your role brief. Under the API terms we use, Anthropic does not train its models on this data. We do not send your data to any AI provider except when you use a feature that needs it.

Services we run on

• Supabase: our database and sign-in.
• Vercel: hosting.
• Anthropic: AI processing, as described above.
• Resend: sending you email (digests, alerts).
• TheirStack: hiring-data lookups. It receives only company names from job listings, never anything about you.
• Public sources we read for company research (job boards, SEC filings, state WARN notices, Glassdoor). We look up companies, never people. The one exception: the exact URL you submit to qualify a role, including any personal parameter already in it (a referral or tracking token), can reach the job board when we read the listing or check whether it’s still open. We strip tracking parameters from a LinkedIn URL when reading its content, but not for the open/closed check, which uses the URL as submitted regardless of source.

What we don’t do

• We do not sell your data.
• We never apply to jobs on your behalf. You always click submit.
• We never show your name or activity to other members. Counts only.

Contact

Questions, or want your data deleted: alison.grippo@gmail.com (until a dedicated support address exists).