New — The State of AI Hiring 2026. Every number sourced→
Enterprise Hiring

How to Screen 200 CVs with AI: A Step-by-Step Tutorial

How to screen 200 CVs with AI, step by step: set criteria, upload in batches, review scores, approve candidates.

How to screen 200 CVs with AI, with eight blocks of 25 showing the eight uploads

TL;DR: How to screen 200 CVs with AI in ZeaHire: set your criteria first, upload in eight batches, and decide each candidate yourself.

  1. Set your criteria.
  2. Create or import the job and set the score weights.
  3. Upload the CVs, up to 25 per batch.
  4. Check processing and review any failed files.
  5. Review the AI scores.
  6. Approve or reject each candidate yourself.
  7. Move the shortlist to interview.

Updated 21 October 2026.

A few hundred CVs arrive for a job order, the client is waiting on a shortlist, and reading every file takes hours that a submission cannot wait for. The problem is the volume, not the recruiter. This tutorial walks through a criteria-first workflow in ZeaHire: set the criteria, upload in batches, review the AI scores, and make each decision yourself.

If your desk runs on job orders like this, the wider picture is in AI screening for staffing agencies.

Key takeaways

  • Screening 200 CVs in ZeaHire is eight uploads, not one. Each upload takes up to 25 files (PDF, DOC, DOCX or TXT, 5 MB each), and an ATS import is the alternative.
  • Scoring follows criteria you set first: per-job weights across skills, experience and education, with thresholds for Recommended, Good Potential and Needs Review.
  • The AI recommendation is advisory. A recruiter approves or rejects. In ZeaHire, candidates wait in Pending Review until a person approves or rejects them.
  • Setting criteria first and keeping a human in the review is the pattern across tutorial-style pages on this topic, including Voicit and HRHQ; periodic auditing comes from Voicit.

Before you start: what you need

Gather five things before you open the product.

  1. A job description for the role you are screening.
  2. A criteria matrix: your must-haves, nice-to-haves and knock-outs, each with a weight. The matrix below maps each tier to the field where it lives in ZeaHire's Score Configuration. Print it and fill it in before you read a single CV.
  3. Your CVs, or an ATS to import from. If you are only practising, use synthetic demo data, never real candidates.
  4. A role that can create jobs. Creating a job needs a role with that permission, so ask your admin if the option is unavailable.
  5. A reviewer. Someone has to approve or reject each candidate, because Step 6 is a human decision.

Screening criteria matrix (printable)

CriterionTierWeightWhere it lives in ZeaHire
(e.g. a named skill the role cannot do without)Must-haveSkills weightRequired skills tags on the job
(e.g. a skill that helps but is not essential)Nice-to-haveSkills weightSkills score.
(e.g. years in a comparable role)Must-have or nice-to-haveExperience weightExperience score
(e.g. a degree or certification)Must-have or nice-to-haveEducation weightEducation score
(e.g. no right to work in the role's country)Knock-outn/aApplied by you at review

Plan for the number: 200 CVs is eight uploads of 25 files. If your ATS already holds the CVs, import them instead through Business Integrations (choose Workday or Kombo as the source system), or use the SuccessFactors and Workday import tabs. Candidate import from JazzHR is not shown in the product interface, so this tutorial does not rely on it.

If your ATS already has its own AI screening, read Bullhorn Amplify's built-in screening for what that tool does. This tutorial covers the volume workflow in ZeaHire, and ZeaHire is one tool among several; for the others, see comparing CV screening software.

Set up the job and the scoring

Step 1: Create or import the job

Create the job manually, or import it from a connected system. ZeaHire's job import supports JazzHR, SAP SuccessFactors and Kombo, and a JobDiva connector is present. A widget route also exists.

Add the required skills tags to the job. These are the must-haves from your criteria matrix, and the scoring uses them.

Step 2: Set the Score Configuration

Open the job's Score Configuration. Set the weights across Skills, Experience and Education so that they reflect your matrix. Then set the three thresholds: Recommended, Good Potential and Needs Review.

As an example, a job might use 60% Skills, 20% Experience and 20% Education, with Recommended at 75% or above, Good Potential at 60% or above and Needs Review at 40% or above. Treat these as example values, not defaults, and read your own from the Score Configuration screen itself.

Upload and parse the CVs

Seven-step screening workflow with step 6, approve or reject each candidate, highlighted as the human decision, and 200 CVs shown as eight uploads of 25

Step 3: Add the CVs

Go to Candidates, then Add Candidate, then Upload Resume. Choose Associate with Job; the association is required. Drop the files into the upload area and select Upload and Parse.

The limits apply per upload:

  • File types: PDF, DOC, DOCX or TXT.
  • Size: 5 MB per file.
  • Count: up to 25 files per upload.

For 200 CVs, that is eight batches of 25. If the CVs already sit in an ATS, use the import route from the section above instead.

Step 4: Watch Resume Processing

Open the Resume Processing tab. It lists each file with its size, the parsing method, a Parsing Status and a Progress indicator, and it refreshes automatically.

Check the Parsing Status column after each batch. Review any file marked Failed before you review scores.

Review the scores and decide

Step 5: Review the AI scores in Pending Review

Open Pending Review. Each row shows:

  • an AI Resume Score with Skills, Experience and Education sub-scores;
  • the skills matched (✓) and missing (✗) against the job;
  • years of experience and roles held;
  • an AI recommendation (based on resume) of Needs Review or Recommended.

The recommendation is advisory. Read the matched and missing skills against the CV for any candidate you are about to act on, and read the Needs Review band by hand rather than skipping it.

Step 6: Approve or reject. The call is yours.

For each candidate, open the row menu (⋮) and choose Approve or Reject Application. New applicants enter the active pool only after a human makes that decision.

In ZeaHire, candidates wait in Pending Review until a person approves or rejects them.

Step 7 (optional): Rank the pool you already have

If you already hold a database of candidates, open the job and go to the Recommended tab. It ranks the existing pool by Requirements Match and Profile Confidence, deterministically, and gives a written rationale and a list of Gaps for each candidate. Choose Associate or Ignore for each one.

Move the shortlist forward

Step 8: Interview the shortlist

Interviewing is a separate stage with its own method; how AI interview software works covers it. In ZeaHire, select Schedule Interview and choose an Autonomous AI interview or a panel interview.

Three other actions exist on the shortlist: Export CSV, Submit to Client and Download PDF.

Decide what to ask before the first interview is booked. Structured questions for the shortlist gives you a bank to draw from.

The answer depends on where the role and the candidates sit. Four points matter here.

  • EU AI Act. AI used in recruitment falls under Regulation (EU) 2024/1689, whose Annex III point 4 lists "Employment, workers' management and access to self-employment" among its high-risk areas (EUR-Lex). Regulation (EU) 2026/1744, in force since 27 July 2026, moved the high-risk obligations for Annex III systems from 2 August 2026 to 2 December 2027 (EUR-Lex). Voicit summarises this and the GDPR point below.
  • GDPR Article 22. The article governs decisions based solely on automated processing (GDPR, Regulation (EU) 2016/679).
  • NYC Local Law 144. The law applies to New York City employers and agencies that use automated employment decision tools, and it includes a notice duty toward candidates (NYC Automated Employment Decision Tools). State-by-state detail is in AI hiring laws by state.
  • Practical fairness. Set criteria before you read any CV, have a person review every approve or reject, audit periodically, and avoid requirements that act as proxies for protected traits. Voicit recommends human supervision and a quarterly bias audit, and HRHQ describes AI as a first-round filter only.

One primary source on bias in CV screening is the NBER résumé audit study by Bertrand and Mullainathan (2004, working paper w9873). In our reading, it is a reason to set criteria before opening any CV and to read the Needs Review band by hand, not a measure of this tool.

This tutorial does not cover bias or fairness reporting, and it makes no promise of either.

FAQ

How long does it take AI to screen 200 CVs?

A 200-CV batch is eight uploads of 25 files, and the Resume Processing tab shows progress for each file, so you can watch each batch finish. We quote no vendor speed figure.

How many CVs can I upload at once?

You can upload up to 25 files per upload, each up to 5 MB, in PDF, DOC, DOCX or TXT. Screening 200 CVs therefore takes eight uploads. You can also import candidates from an ATS.

What if a CV won't parse?

The Resume Processing tab marks it with a Failed status.

It depends on where the role and the candidates are. The EU AI Act, GDPR Article 22 and New York City's Local Law 144 are the main references; the legal section above sets them out.

Will AI reject good candidates?

The score is a recommendation, and a person approves or rejects. Read the Needs Review band by hand.

How do I avoid bias in AI CV screening?

Set criteria before reading CVs, have a person review every decision, and audit periodically. Avoid proxy requirements. This tutorial does not promise bias or fairness reports.

Should I tell candidates AI is used?

Disclosure requirements depend on the jurisdiction. NYC Local Law 144 includes candidate notice (NYC DCWP). Check the rules where the role and the candidates sit; AI hiring laws by state tracks them.

Can I import candidates from my ATS instead?

Yes, for some systems. Business Integrations (Workday or Kombo as the source) and the SuccessFactors and Workday import tabs exist. Candidate import from JazzHR is not shown in the interface, so confirm before relying on it.

Can a recruiter run this without an admin?

That depends on the permissions your organisation assigns. Ask your admin whether your role can create jobs and schedule interviews.

Next step

See an evidence pack in a demo to see what a scored candidate record looks like.

Hiring processRecruiting
Was this useful?

Comments

No comments yet — start the conversation.

Leave a comment

Your email is never published. Comments appear once a moderator approves them.

Replying to
Plain text only. Be kind and stay on topic.

By commenting you agree to our privacy policy.