For HR software and recruiting teams
AI resume screening with scores recruiters can check
Parse each application, rank up to 10 candidates against the job in one call, then open any score to see the requirement it rests on. HireLayer gives your screening feature its evidence; the decision stays with the recruiter.
The problem
Why screening is hard to get right
Most teams already screen with filters and a fast read. The trouble starts when volume grows or someone asks why a candidate was set aside.
- 01
The first applications get the attention
With a few hundred CVs per role, recruiters read the top of the list closely and skim the rest.
- 02
Keyword filters miss equivalent experience
“Led a support team of six” does not contain “people management”, so a filter drops a candidate a recruiter would keep.
- 03
A bare score does not survive a question
A hiring manager asks why one profile ranks above another. “The tool said 0.82” is not an answer.
- 04
Regulators expect explanations
The EU AI Act lists filtering applications and evaluating candidates as high-risk. Screening needs reasons a person can review.
How it fits
From application to reviewed shortlist
Your product keeps the jobs, the applications and the decision. HireLayer structures each document and returns scores with their reasons.
- Your system
Step 1, your system: Applications arrive
Your ATS or careers page receives the CV files for a job, as it does today.
- HireLayer
Step 2, HireLayer API: Parse each CV
One call per file returns the candidate record and the full CV text used for screening.
POST /
api/ v3/ parser - info_resume.text
- work_experiences[]
- skills[]
- HireLayer
Step 3, HireLayer API: Turn the job into criteria
Once per job, the description becomes weighted requirements with a mandatory flag. A recruiter adjusts them before screening starts.
POST /
api/ v1/ jobs/ extract-criteria - matching_criteria[]
- HireLayer
Step 4, HireLayer API: Rank the pool
Send up to 10 candidates for the job in one request. Each comes back once, with a rank, a score from 0 to 1 and a rationale.
POST /
api/ v1/ matching/ job-candidates/ rank - rankings[]
- HireLayer
Step 5, HireLayer API: Explain the finalists
For the profiles a recruiter opens, evaluate every criterion: ideal, potential, not_mentioned or not_valid, each with the evidence from the CV.
POST /
api/ v1/ matching/ job-candidate - score
- evaluated_criteria[]
- Your system
Step 6, your system: The recruiter decides
Your interface shows the order and the reasons. Advancing or rejecting a candidate remains a person’s action, logged in your product.
Want the whole flow as a program you can run on your own CVs?
Read the screening guideScreening queue
A ranked queue where every score opens up
What a screening view in your product can show with Rank and Match results. Rationales and explanations come from the API, in French in the current version; the decision buttons are yours.
Technical example
Rank three applications for one job
Rank needs only the job text and the CV texts. Send info_resume.text from CV Extract, up to 50,000 characters each.
- candidate_id
- Your application id, returned unchanged.
- score
- From 0 to 1, comparable within this request only.
- rationale
- Why the candidate holds this position, in French.
curl -X POST https://hirelayer.co/api/v1/matching/job-candidates/rank \
-H "Content-Type: application/json" \
-H "X-API-Key: YOUR_API_KEY" \
-d '{
"job_text": "Customer Success Manager. 3+ years with B2B SaaS accounts. Fluent English. Hybrid, Lyon.",
"candidates": [
{ "id": "app_1041", "candidate_text": "<info_resume.text>" },
{ "id": "app_1042", "candidate_text": "<info_resume.text>" },
{ "id": "app_1043", "candidate_text": "<info_resume.text>" }
]
}'{
"rankings": [
{
"rank": 1,
"candidate_id": "app_1042",
"score": 0.86,
"rationale": "Quatre ans de gestion de comptes SaaS B2B et un anglais courant attesté par un poste à Londres."
},
{
"rank": 2,
"candidate_id": "app_1041",
"score": 0.71,
"rationale": "Solide expérience client en SaaS, mais le niveau d'anglais n'est pas précisé."
},
{
"rank": 3,
"candidate_id": "app_1043",
"score": 0.48,
"rationale": "Profil support technique, peu d'éléments sur la gestion de comptes."
}
]
}APIs used
The HireLayer APIs behind this workflow
Start with the core APIs, add the others when a feature needs them. All five share one API key and one credit balance.
- HireLayer Rank
Candidate Ranking API
Core to this workflowDocs for HireLayer RankOrders up to 10 applications for a job, each with a score and a rationale.
- HireLayer Match
Candidate & Job Matching API
Core to this workflowDocs for HireLayer MatchExplains a candidate requirement by requirement, with the evidence from the CV.
- HireLayer CV Extract
Resume Parsing API
Core to this workflowDocs for HireLayer CV ExtractTurns each CV file into a record and the text used for screening.
- HireLayer Job Extract
Job Parsing API
Add when neededDocs for HireLayer Job ExtractTurns the job description into weighted criteria a recruiter can edit.
- HireLayer Skills
Skills Matching API
Add when neededDocs for HireLayer SkillsNormalizes skill labels so filters and search agree across CVs.
Outcomes
What changes for recruiters and your product
Every application gets read
The pool is ranked against the same job, so a strong CV at the bottom of the inbox is not missed.
Equivalent experience counts
Requirements are evaluated from what the CV describes, not from a list of exact keywords.
Scores that can be explained
Each status points to a sentence in the CV, so a recruiter can answer a hiring manager or a candidate.
A person stays in charge
HireLayer returns an order and reasons. Your product records who advanced or rejected each candidate.
FAQ
AI resume screening: frequently asked questions
What is AI resume screening?
AI resume screening reads the CVs submitted for a job and orders them by how well they meet its requirements, so recruiters review the strongest applications first. With HireLayer, each score comes with a rationale and, on request, a status and an explanation for every requirement.
Does HireLayer reject candidates automatically?
No. The APIs return ranks, scores and explanations. Whether a candidate advances is decided in your product, by a person. The mandatory flag on a criterion is informational and never removes a candidate on its own.
How many applications can be screened at once?
Rank compares up to 10 candidates in one request. For larger pools, pre-filter with your own rules first, or score each application with Match, which evaluates one candidate per call against the job criteria.
Is AI resume screening high-risk under the EU AI Act?
Yes. Annex III of the AI Act lists AI used to filter applications and evaluate candidates as high-risk. Under the Digital Omnibus, those rules apply from 2 December 2027. They call for human oversight and records of how the system was used, which per-criterion explanations and request ids help you document.
How much does screening one job cost?
One credit per successful call. Screening 10 applications typically takes 1 credit for the job criteria, 10 for parsing, 1 for the ranking and 1 per finalist explained with Match: 15 credits with three finalists. The free plan includes 50 credits a month.
Can recruiters screen candidates from ChatGPT or Claude?
Yes. The HireLayer MCP server exposes parsing, matching and ranking as tools in Claude, ChatGPT, Cursor and VS Code, so a recruiter can rank a shortlist in a conversation without an integration project.
What language are rationales and explanations written in?
Rank rationales, Match summaries and explanations, and Job Extract criteria are generated in French in the current version. CVs and jobs can be in other languages.
Screen one real job with it
Take a job you are hiring for, parse its applications and compare the order with your recruiters’ own. The free plan includes 50 credits a month, shared by all five APIs.