Infrastructure for Recruitment Candidate Screening

Candidate screening AI requires documented job requirements and qualification criteria — minimum education, required licenses, years of experience — to automate disqualification. Healthcare organizations have these documented in job descriptions, though the baseline notes they're rarely updated. For automated screening of required qualifications (RN license, BLS certification), existing documentation practices suffice at L2. However, undocumented success criteria and informal hiring manager preferences limit how far screening automation can extend without human review.

The screening AI requires systematic capture of candidate applications through the ATS, structured interview scores, hiring decisions, and subsequent performance outcomes of hired candidates. Without template-required fields capturing who was hired, interview scores, and 90-day performance ratings, the AI cannot train predictive models. The ATS systematically captures applications; the gap is structured capture of interview outcomes and post-hire performance linked back to candidate records.

Resume screening and candidate scoring require consistent schema: Candidate records with fields for education, licensure, experience years, ATS score, interview score, and performance rating. All records must share these fields for the AI to compute predictive rankings. Job records must specify required qualifications as structured criteria, not free-text descriptions. Consistent schema enables the AI to compute 'Candidate.License = Required.License' rather than parsing unstructured job descriptions.

Candidate screening operates primarily within the ATS — the AI needs access to candidate applications, job requirements, and historical hiring outcomes within that system. Current point integrations between ATS and HRIS (for hired candidate data) partially support this. At L2, the AI can screen within the ATS context but lacks real-time access to performance management data for training predictive models. Manual export of post-hire performance data to train screening models is acceptable for this use case.

Job requirements for healthcare roles change when licensing requirements change (state board rule updates) or when clinical protocols shift required competencies. At L2, periodic review of job descriptions and screening criteria is sufficient — most screening logic (RN license required, BLS required) remains stable between reviews. The baseline notes job descriptions are rarely updated, which constrains predictive screening quality but is acceptable for basic qualification filtering with scheduled updates.

Recruitment screening requires the ATS to connect to HRIS (for hired candidate tracking and performance outcome lookup) and the job requirements database. Point-to-point integration between ATS and HRIS supports the core use case — matching candidates to roles and tracking outcomes. Deeper integration to clinical quality systems or credentialing isn't required at the screening stage. The baseline confirms ATS and HRIS exist and integrate at a basic level sufficient for this capability.