# Sightline Hire — Full Marketing Content > Sightline Hire is a product by HR Maximizer. It is fair, explainable AI-driven candidate review software for HR teams: a stack of resumes becomes a ranked, evidenced shortlist in minutes. --- ## Hero **See every candidate clearly. Hire with confidence.** Fast, fair, explainable AI-driven candidate review — built on speed, accuracy, fairness, and trust. Trust pills: SOC 2 aligned · AES-256 encrypted · Privacy by default. --- ## Who it's for One tool, three jobs done well. - **Recruiters** — Triage hundreds of inbound resumes per role into a ranked, evidenced shortlist — and draft personalized follow-ups in the same place. - **Hiring managers** — Get a Tier A / B / C / D view with per-criterion findings you can defend in a debrief — instead of scanning resumes between meetings. - **HR ops & compliance** — Identity is masked by default and every reveal is audited. Show a clean access log on demand and keep retention defaults aligned to policy. --- ## How it works From stack of resumes to shortlist — in minutes. A focused workflow designed for recruiters and hiring managers who care about both throughput and fairness. 1. **Upload the job & resumes.** Paste a JD or link, drop in up to 300 resumes per role. PDF, DOCX, TXT — we sanitize and de-identify before any AI sees them. 2. **AI evaluates against your criteria.** Each resume is scored on every criterion you confirm — with verbatim evidence quotes, not vibes. Star-resume calibration anchors keep the bar honest. 3. **Shortlist with confidence.** Tiered results with explainable findings, follow-up questions, and one-click drafts. Reveal names only when you're ready — every reveal is audited. --- ## Capabilities (what you get) Six capabilities that turn a stack of resumes into a ranked, evidenced shortlist — without the shortcuts that get teams in trouble. ### Scoring with evidence — Every score is backed by a verbatim quote. We score each resume against the criteria you confirm — never a fuzzy "overall fit" number. Each criterion gets a rating, a short rationale, and the exact line from the resume that supports it. - Per-criterion ratings, not a single black-box score - Verbatim evidence quotes — no hallucinated credentials - Click any finding to jump to the source in the resume ### Tiered shortlists — Tier A through D — ranked, not just scored. Candidates land in clear tiers (A: strong match, B: promising, C: gaps, D: not a fit) so your hiring team can triage in a glance. Within each tier, candidates are ranked by weighted criterion coverage. - Tier A / B / C / D with explainable thresholds - Ranked inside each tier, with reasons you can defend - Re-evaluate any time you adjust criteria or star resumes ### Star-resume calibration — Anchor the bar to people you've already hired well. Upload up to 5 resumes of top performers per role. We de-identify them, summarize what made them strong on each criterion, and use them as calibration anchors — never as a shortcut for credit-by-association. - Capture institutional knowledge from your best hires - Strict prompt rules block "looks like" favoritism - Stars are de-identified before any model sees them ### Privacy-by-default identity — Names stay hidden until you choose to reveal them. Lists, dashboards, leaderboards, and exports all show a sanitized code like `resume-A7K2.pdf` — not the candidate's name. Recruiters reveal a name only when they need to, and every reveal is logged with who, when, and why. - Default masking across every screen and admin view - One-click reveal — individual or bulk — when warranted - Per-candidate access log you can show to compliance ### Follow-up email drafts — From shortlist to outreach without retyping a thing. For every candidate, we draft a personalized follow-up email keyed to the gaps and strengths the evaluator found — with targeted clarifying questions you can edit, copy, or open in your mail client. - Subject + body pre-filled from per-criterion findings - Bulk personalized outreach for an entire tier - Your email signature is remembered across sessions ### Fairer by construction — A workflow designed to remove the wrong shortcuts. Identity is masked before evaluation. Star resumes can only inform the bar, never grant credit. Prompt-injection patterns in resumes are detected and surfaced. Every decision leaves a trail you can audit later. - Names, photos, and demographic cues removed pre-evaluation - Injection-defense sanitizer flags suspicious resume text - Immutable audit events for every reveal and override --- ## Outcomes - **300** — Resumes per job, per pass. Upload a full applicant pool — up to 300 candidates — and get a ranked, evidenced shortlist in minutes. - **Per-criterion** — Scoring, not gut feel. Replace vague "looks like a fit" with structured ratings and quoted evidence for every requirement you set. - **100%** — Auditable decisions. Every evaluation, override, and identity reveal is captured — defensible for compliance, fair for candidates. --- ## Security & privacy, built in Sightline Hire is designed around the SOC 2 Trust Services Criteria and modern data-protection regulations (GDPR, CCPA, PIPEDA). Privacy isn't a setting — it's the default. - **Privacy-by-default identity** — Candidate names are masked in every list and dashboard. Recruiters explicitly reveal a name when they need it — and every reveal is audited. - **Encrypted end-to-end** — HTTPS in transit, AES-256 at rest, OIDC sign-in with PKCE, and per-user authorization on every API route. - **AI prompt-injection defenses** — Resumes are sanitized before they reach the evaluator. Suspicious patterns are flagged so recruiters review with full context. - **Explainable & auditable** — Every evaluation cites per-criterion findings. Every identity reveal writes an immutable audit event you can inspect. --- ## FAQ **What AI model do you use, and can it see candidate names?** Evaluations run on GPT-5.4 with strict JSON output. Resumes are sanitized first — names, contact info, photos, and obvious demographic cues are stripped before any model sees the text. The model never sees the candidate's real identity, and prompt-injection patterns in resumes are flagged before evaluation. **Where is our data stored, and is it used to train AI?** All resumes, evaluations, and audit events live in your private workspace database, encrypted with AES-256 at rest. Files are stored in object storage with signed-URL access. We do not use your data to train models — calls to the AI provider run under no-training terms. **How long do you keep raw resumes?** Default retention for uploaded files is 30 days, configurable per candidate up to a hard 180-day cap. Evaluation results and audit events are kept for the life of the workspace so your defensible decision trail stays intact. **Can I export results to share with a hiring panel?** Yes. Shortlists and per-candidate evaluations export with their tier, per-criterion findings, and the verbatim evidence quotes. Candidate names stay masked unless a recruiter has explicitly revealed them — and that reveal is logged. **How is this different from a black-box resume screener?** We don't return a single "fit score". Every rating is per-criterion, every finding cites a verbatim quote, every reveal is audited, and star-resume calibration is bounded by strict prompt rules that forbid credit-by-association. The point is a decision you can defend in writing. **Do candidates know they're being evaluated by AI?** That's your call and depends on your jurisdiction. We give you the audit trail (decisions, reveals, overrides) and per-criterion explanations so you can disclose accurately — including under laws like NYC LL 144, EU AI Act, and Illinois AIVI. --- ## Operational facts - Upload limits: **300 candidates per job per pass**, **15 MB per file**, formats **PDF / DOCX / TXT**. - Retention: raw files default **30 days**, configurable up to **180 days** max. Evaluations and audit events retained for the workspace lifetime. - Pipeline: parse (pdf-parse / mammoth) → AI sanitize (strip PII, neutralize prompt injection) → AI evaluate against confirmed criteria. Concurrency 2 per upload batch. - AI: model **GPT-5.4**, `response_format: json_object`. - Auth: Replit Auth via OIDC with PKCE, session cookies, per-user authorization on every API route. --- ## Brand - **Product name:** Sightline Hire - **Maker:** HR Maximizer - **Tagline:** See every candidate clearly. Hire with confidence. - **Marketing site:** https://sightlinehire.com/welcome - **Request a demo / pricing:** https://sightlinehire.com/welcome#request-demo