W K
EU AI Act III(4)(a): High Risk Q3

Candidate Screening Agent

Structure the screening process - with full EU AI Act compliance built in.

Analyses applications against requirement profiles and prepares structured shortlists. EU AI Act high-risk system with enhanced governance.

Score Dashboard

Agent Readiness 64-71%
Governance Complexity 74-81%
Economic Impact 78-85%
Lighthouse Effect 76-83%
Implementation Complexity 51-58%
Transaction Volume Daily

What This Agent Does

Candidate screening is the most governance-intensive recruiting process. Every application must be evaluated fairly, consistently, and traceably - and under the EU AI Act, any AI system used in recruiting is classified as high-risk (Annex III, Section 4(a)). This means the governance bar is not optional: it is regulatory. The Candidate Screening Agent does not make screening decisions. It structures the process. The agent parses incoming applications (CVs, cover letters, structured application forms), extracts qualifications and experience against the job requirement profile, identifies potential matches and gaps, and presents this structured view to the recruiter who makes the screening decision. Every step is logged: what data was extracted, which requirements were matched, what the confidence level was, and what the recruiter decided. This audit trail is not just good practice - it is a legal requirement under the EU AI Act for high-risk systems. The agent also monitors for bias indicators in its own outputs: if certain demographic groups are systematically ranked lower or filtered out, the monitoring system flags this for review. Bias monitoring is continuous, not a one-time check. This is a Q3 agent. The governance infrastructure required (decision logging, rule versioning, human-in-the-loop patterns, bias monitoring) should be built and proven in Q1 agents before being applied here. Starting with candidate screening before the governance infrastructure exists is the most common deployment mistake organisations make.

Micro-Decision Table

Human
Rules Engine
AI Agent
Each row is a decision. Expand to see the decision record and whether it can be challenged.
Parse application documents Extract structured data from CV, cover letter, application form AI Agent

Document parsing and data extraction from varied formats

Decision Record

Model version and confidence score
Input data and classification result
Decision rationale (explainability)
Audit trail with full traceability

Challengeable: Yes - fully documented, reviewable by humans, objection via formal process.

Match qualifications to requirements Compare extracted qualifications against job requirement profile AI Agent

AI-assisted matching with confidence scores per requirement

Decision Record

Model version and confidence score
Input data and classification result
Decision rationale (explainability)
Audit trail with full traceability

Challengeable: Yes - fully documented, reviewable by humans, objection via formal process.

Identify qualification gaps Flag requirements not clearly met by application data AI Agent

Gap analysis comparing profile to requirement list

Decision Record

Model version and confidence score
Input data and classification result
Decision rationale (explainability)
Audit trail with full traceability

Challengeable: Yes - fully documented, reviewable by humans, objection via formal process.

Generate structured candidate profile Present extracted data and match assessment to recruiter AI Agent

Automated profile assembly for consistent recruiter review

Decision Record

Model version and confidence score
Input data and classification result
Decision rationale (explainability)
Audit trail with full traceability

Challengeable: Yes - fully documented, reviewable by humans, objection via formal process.

Run bias monitoring check Analyse output distribution for demographic bias indicators AI Agent

Statistical fairness analysis on screening outputs

Decision Record

Model version and confidence score
Input data and classification result
Decision rationale (explainability)
Audit trail with full traceability

Challengeable: Yes - fully documented, reviewable by humans, objection via formal process.

Flag bias concern Alert compliance team if bias indicators exceed threshold Rules Engine

Threshold-based alerting per defined fairness metrics

Decision Record

Rule ID and version number
Input data that triggered the rule
Calculation result and applied formula

Challengeable: Yes - rule application verifiable. Objection possible for incorrect data or wrong rule version.

Recruiter reviews profile Evaluate candidate based on structured profile and own assessment Human

Human decision required - AI provides structure, not verdict

Decision Record

Decider ID and role
Decision rationale
Timestamp and context

Challengeable: Yes - via manager, works council, or formal objection process.

Record screening decision Document recruiter's decision with reasoning Human

Mandatory documentation per EU AI Act audit trail requirement

Decision Record

Decider ID and role
Decision rationale
Timestamp and context

Challengeable: Yes - via manager, works council, or formal objection process.

Log decision for audit trail Store complete decision record with all inputs and outputs Rules Engine

Automated logging per high-risk system documentation requirements

Decision Record

Rule ID and version number
Input data that triggered the rule
Calculation result and applied formula

Challengeable: Yes - rule application verifiable. Objection possible for incorrect data or wrong rule version.

Update candidate status Move candidate to next stage or rejection Rules Engine

Status update based on recorded recruiter decision

Decision Record

Rule ID and version number
Input data that triggered the rule
Calculation result and applied formula

Challengeable: Yes - rule application verifiable. Objection possible for incorrect data or wrong rule version.

Generate rejection documentation Produce GDPR-compliant rejection notification if applicable AI Agent

Automated notification generation per configured templates

Decision Record

Model version and confidence score
Input data and classification result
Decision rationale (explainability)
Audit trail with full traceability

Challengeable: Yes - fully documented, reviewable by humans, objection via formal process.

Decision Record and Right to Challenge

Every decision this agent makes or prepares is documented in a complete decision record. Affected employees can review, understand, and challenge every individual decision.

Which rule in which version was applied?
What data was the decision based on?
Who (human, rules engine, or AI) decided - and why?
How can the affected person file an objection?
How the Decision Layer enforces this architecturally →

Prerequisites

  • Applicant tracking system (ATS) with structured data model
  • Job requirement profiles with measurable qualification criteria
  • EU AI Act conformity assessment documentation
  • Bias monitoring framework with defined fairness metrics
  • Decision logging infrastructure with full audit trail
  • Works council agreement on AI-supported screening (Art. 26(7) EU AI Act)
  • Data Protection Impact Assessment for automated candidate processing
  • Fundamental rights impact assessment per EU AI Act requirements
  • Human-in-the-loop workflow ensuring no automated screening decisions

Governance Notes

EU AI Act III(4)(a): High Risk
Classified as high-risk under the EU AI Act, Annex III, Section 4(a) - AI systems intended for use in recruiting or selection of candidates. Full conformity assessment mandatory before deployment. Article 26(7) requires informing worker representatives before introducing the system. Continuous bias monitoring is required, not optional. GDPR Articles 13-14 require informing candidates about automated processing. Article 22 right to not be subject to solely automated decisions applies - the agent must ensure a human makes every screening decision. A fundamental rights impact assessment must be completed. The agent's governance requirements are the strictest in this catalog. The Decision Layer decomposes every process into individual decision steps and defines for each: Human, Rules Engine, or AI Agent. Every decision is documented in a complete decision record. Affected employees can understand and challenge any automated decision.

Infrastructure Contribution

The Candidate Screening Agent is the litmus test for high-risk governance readiness. If an organisation can deploy this agent with full EU AI Act compliance - conformity assessment, bias monitoring, decision logging, human-in-the-loop - it can deploy any high-risk agent. The governance infrastructure validated here directly transfers to the Performance Review Documentation Agent, Merit Cycle Governance Agent, and Promotion Process Agent. Builds Decision Logging and Audit Trail used by the Decision Layer for traceability and challengeability of every decision.

Related Pages

Decision Layer
Governance layer between AI agent and target system. Every decision documented and challengeable.
EU AI Act
High-risk requirements: risk management, transparency, human oversight.
Co-Determination & AI
Works council, works agreements, and co-determination as architecture constraint.
HR AI Agents
All HR agent services at a glance.

Detailed Analysis Available

A deep-dive for Candidate Screening Agent is available with complete micro-decision decomposition, industry variants, and implementation details.

Read Deep-Dive

Related Agents

Job Posting Agent

Publish compliant, consistent job postings - across every channel, every language.

Q3
D K
EU AI Act III(4)(a): High Risk
Readiness: 71-78%
Economic: 56-63%
Governance: 58-65%
Micro-Decisions: 8
Weekly

Interview Scheduling Agent

Find the slot, book the room, send the invite - without the email ping-pong.

Q2
W
Readiness: 84-91%
Economic: 51-58%
Governance: 6-13%
Micro-Decisions: 6
Daily

Pre-Hire Due Diligence Agent

Structured background verification - legally compliant, consistently documented.

Q3
D W
EU AI Act III(4)(a): High Risk
Readiness: 58-65%
Economic: 51-58%
Governance: 78-85%
Micro-Decisions: 9
Weekly
Back to Catalog

Frequently Asked Questions

Does the agent reject candidates automatically?

No. The agent structures information for human review. Every screening decision is made by a recruiter. The EU AI Act prohibits fully automated decisions in recruiting - this agent is designed with human-in-the-loop as a fundamental architectural requirement.

Why is this a Q3 agent and not Q1?

The governance requirements for high-risk AI in recruiting are the most demanding in the catalog. Decision logging, bias monitoring, conformity assessment, and works council agreement must be in place before deployment. These capabilities are built and proven in Q1 agents (payroll, time management) first.

How does the bias monitoring work?

The agent continuously analyses its output distribution across demographic groups (where permitted by law). If systematic disparities emerge - for example, candidates from certain backgrounds consistently receiving lower match scores - the system flags this for compliance review. Monitoring is statistical and ongoing, not a one-time certification.

Implement This Agent?

We assess your process landscape and show how this agent fits into your infrastructure.