Skip to content
    Charviam.We build. You grow.
    HR & Recruiting

    Job Application Screener

    Evidence-linked summaries; recruiters make every decision.

    Always human-reviewed (regulated)Content & drafts

    Job Application Screener

    HR & Recruiting

    Draft mode
    IllustrativeNot live data
    1. RuleDone:

      A new application arrives in your ATS.

    2. RuleDone:

      It checks for an approved rubric, signed attestation, legal review and required notices.

    3. RuleDone:

      It reads the resume and removes protected and personal details.

    4. RuleDone:

      It finds quoted evidence for each criterion, with no score.

    5. PersonWaiting for approval:Waiting for a person

      The recruiter reviews the card and accepts the private note.

    6. RuleNot started:

      Weekly: a selection-rate export for HR and auditors.

    For example: A 200-person services firm hiring an electrician gets 300 applications in Greenhouse.

    Waiting for a person. Nothing has changed yet.

    Always human-reviewed (regulated). A person decides. It never acts on its own.

    What is Job Application Screener?

    In short

    Job Application Screener is an AI automation for HR & Recruiting teams that works with Gmail, Greenhouse, Lever, Outlook, Workable and Zoho Recruit. Summarises each application against your own job rubric with quoted evidence, so recruiters review faster. It never scores, ranks, rejects or advances anyone. It is built for head of talent / HR (signs the attestation), recruiters, hiring managers and HR admin.

    Last reviewed

    Key facts

    Department
    HR & Recruiting
    Starts when
    A new application is submitted in your ATS for a job with an approved rubric
    Works with
    Gmail, Greenhouse, Lever, Outlook, Workable, Zoho Recruit
    Run modes offered
    Shadow and Draft
    Human control
    Always human-reviewed (regulated)
    Status
    Available now
    Before

    The problem today

    Each role gets 100–1,000 applications. Recruiters skim CVs in 30–60 seconds each, inconsistently, and miss qualified people. Notes are not tied to the job requirements, so decisions are hard to explain later.

    With Job Application Screener

    What it does

    When a new application arrives for a job with an approved rubric, it reads the resume and removes names, contact details, age signals, school names and other protected details. It then finds evidence for each criterion (met, partly or not found) with quotes. A short neutral summary and phone-screen questions go on a review card for the job's recruiter, and if the recruiter accepts, a private note is added in the ATS. A weekly selection-rate export helps HR and an independent auditor check for adverse impact.

    How does Job Application Screener work?

    How it works
    • Rule you set
    • AI step
    • A person decides
    1. Step 1Rule

      A new application arrives in your ATS.

    2. Step 2Rule

      It checks for an approved rubric, signed attestation, legal review and required notices.

    3. Step 3Rule

      It reads the resume and removes protected and personal details.

    4. Step 4Rule

      It finds quoted evidence for each criterion, with no score.

    5. Step 5Person

      The recruiter reviews the card and accepts the private note.

    6. Step 6Rule

      Weekly: a selection-rate export for HR and auditors.

    What you get

    Key features
    • Status per criterion (met / partly / not found) with quotes; a missing must-have reads "not found in the documents: check in screen".
    • Redacts name, contact details, age and graduation years, gender words, nationality, religion, caste, disability, health, school names and career-gap reasons.
    • Rubric word list blocks "culture fit", "native speaker", "recent graduate" and similar criteria.
    • A neutral banner for hidden or instruction-like text in a resume, never a negative mark.
    • Jurisdiction notices (NYC, Colorado, Illinois, EU), including the NYC 10-business-day lead time.
    • Four-fifths rule impact ratios per category; groups under 30 flagged as unreliable.

    How much does it do on its own?

    Run modes

    Pick a mode to see what Job Application Screener would do at each step.

    Every automation starts in Shadow. Move it up when you are ready, and back down at any time. How run modes work

    Job Application Screener in Shadow

    Runs on real data and shows what it would have done. No app is changed and no message is sent.

    1. Rule

      A new application arrives in your ATS.

      Watches for it

    2. Rule

      It checks for an approved rubric, signed attestation, legal review and required notices.

      Runs on real data, read only

    3. Rule

      It reads the resume and removes protected and personal details.

      Runs on real data, read only

    4. Rule

      It finds quoted evidence for each criterion, with no score.

      Runs on real data, read only

    5. Person

      The recruiter reviews the card and accepts the private note.

      Shown to you as a suggestion

    6. Rule

      Weekly: a selection-rate export for HR and auditors.

      Shown to you as a suggestion

    Not offered: Assisted, Auto. Hiring and outbound calling: a person always decides.

    Shadow vs Draft for Job Application Screener

    Draft is the most it can do on its own, because Auto is not offered for it. Here is each step side by side.

    What each step of Job Application Screener does in Shadow mode and in Draft mode
    StepShadowDraft
    In shortRuns on real data and shows what it would have done. No app is changed and no message is sent.Every action waits as an approval card you can approve, edit or reject.
    Step 1A new application arrives in your ATS.Watches for itPrepares it
    Step 2It checks for an approved rubric, signed attestation, legal review and required notices.Runs on real data, read onlyPrepares it
    Step 3It reads the resume and removes protected and personal details.Runs on real data, read onlyPrepares it
    Step 4It finds quoted evidence for each criterion, with no score.Runs on real data, read onlyPrepares it
    Step 5The recruiter reviews the card and accepts the private note.Shown to you as a suggestionWaits as an approval card
    Step 6Weekly: a selection-rate export for HR and auditors.Shown to you as a suggestionOnly after you approve
    True in every modeShadow and Draft only, forever. It never moves a stage, rejects, advances, archives or sends rejection emails, and its credentials cannot.

    What it will never do on its own

    Safety & human control
    • Shadow and Draft only, forever. It never moves a stage, rejects, advances, archives or sends rejection emails, and its credentials cannot.
    • There is no overall score, rank, fit label or recommendation in the output.
    • EEO self-ID data never reaches AI or recruiters; only group counts reach the export.
    • What was sent to AI is scanned afterwards; any protected term voids the run and alerts ops and HR.
    • Candidates who ask for accommodation or human-only review get no AI processing; raw resumes are never stored.
    • You approve before it acts
    • Your data stays yours
    • It knows its limits
    • Stop anytime
    How these promises work

    What will you see on your dashboard?

    Results you can track

    Targets are quality goals the automation is built to reach. Estimates are calculated from your own volumes.

    • Target

      Recruiter time per application (target: from ~8 minutes to under 3)

      A goal it is built to reach

    • Tracked

      Time to first review

      Measured on your own data

    • Tracked

      Applications held for notice or sent to manual review

      Measured on your own data

    • Target

      Criterion agreement trend (target: ≥ 0.85 recruiter–rubric agreement)

      A goal it is built to reach

    • Tracked

      Selection rates and impact ratios per category (HR-only view, never per candidate)

      Measured on your own data

    • Target

      Evidence grounding (target: 100%)

      A goal it is built to reach

    • Target

      Protected-term leaks (target: 0)

      A goal it is built to reach

    • Target

      Counterfactual consistency (target: ≥ 98%)

      A goal it is built to reach

    • Target

      Impact ratio on the test set (target: ≥ 0.80)

      A goal it is built to reach

    Illustrative example
    A 200-person services firm hiring an electrician gets 300 applications in Greenhouse. For each one, the recruiter sees "Licence: met" with the quoted line, "Commercial wiring: not found in the documents: check in screen", and three phone-screen questions. She makes every decision in Greenhouse herself.
    A made-up scenario to show how it works, not a customer story.

    Who is Job Application Screener for?

    • Head of talent / HR (signs the attestation)
    • Recruiters
    • Hiring managers
    • HR admin

    Industries: Recruiting & staffing, Professional services, SaaS & tech, Manufacturing & distribution

    Also useful for: Legal & Compliance

    Connects to the apps you already use

    Works with

    App names are trademarks of their owners. Charviam connects to them; it is not a partner or reseller.

    • Gmail(optional)
    • Greenhouse
    • Lever
    • Outlook(optional)
    • Workable
    • Zoho Recruit

    What do you need to set it up?

    Setup
    • Sign the client attestation (six confirmations, renewed yearly) and record your counsel's review.
    • Connect Greenhouse, Lever, Workable or Zoho Recruit (read + note only), plus a separate EEOC-only connection for the export (Greenhouse / Lever).
    • Write and HR-approve a job-related rubric for each job.
    • Choose jurisdictions, counsel-approved notice templates and the notice channel.
    • Set retention (default 3 years) and audit categories; have a bias-audit plan (NYC: a named independent auditor).

    Typical setup

    1. 01Connect your apps
    2. 02Configure the rules
    3. 03Run a watch-only trial
    4. 04Go live

    Not included

    • Rejecting, advancing, ranking, stage changes or rejection emails.
    • Video, voice, facial or personality analysis, and social media scraping.
    • Background checks and salary history.

    Questions about Job Application Screener

    FAQ

    Not fully. It offers Shadow, Draft only. Always human-reviewed.

    Sign the client attestation (six confirmations, renewed yearly) and record your counsel's review. Connect Greenhouse, Lever, Workable or Zoho Recruit (read + note only), plus a separate EEOC-only connection for the export (Greenhouse / Lever). Write and HR-approve a job-related rubric for each job. Choose jurisdictions, counsel-approved notice templates and the notice channel. Set retention (default 3 years) and audit categories; have a bias-audit plan (NYC: a named independent auditor).

    Rejecting, advancing, ranking, stage changes or rejection emails. Video, voice, facial or personality analysis, and social media scraping. Background checks and salary history.

    Shadow and Draft only, forever. It never moves a stage, rejects, advances, archives or sends rejection emails, and its credentials cannot. There is no overall score, rank, fit label or recommendation in the output. EEO self-ID data never reaches AI or recruiters; only group counts reach the export. What was sent to AI is scanned afterwards; any protected term voids the run and alerts ops and HR. Candidates who ask for accommodation or human-only review get no AI processing; raw resumes are never stored.

    See Job Application Screener on your own data

    Walkthrough

    It starts in Shadow, where it only watches and suggests, so nothing changes in your apps until you say so.