The agentic workflow — five agents, one audit trail
Intent classification, pronoun resolution, turn routing with confidence thresholds
Parses job descriptions into structured rubrics, validates skills
pgvector retrieval, rubric scoring, gap & red-flag detection
Generates tailored technical questions & outreach drafts
Halts email sends and calendar locks pending explicit human confirm
What automation actually does
Before RecruitAI (manual)
- Circadian fatigue reading 200 PDFs
- Spreadsheet scoring with ±30% variance across interviewers
- Generic mass-blast emails with <5% reply rate
- No traceability for compliance or bias audits
With RecruitAI (automated)
- Semantic ranking in minutes, same rubric for all candidates
- Deterministic scorecards with evidence citations
- Personalized outreach drafts citing projects — 1-click approve
- Timestamped audit log + bias-impact reporting ready for review
FAQ — Recruitment automation
Does automation replace recruiters?▾
No. RecruitAI automates the repetitive 80% (parsing, scoring, draft generation) and reserves the 20% that requires judgment (final interview, offer decisions, reveals) for humans. Every outbound action is gated behind explicit confirmation.
What agents handle each step?▾
Supervisor routes intent → JD Agent structures job requirements → Screening/RAG agent performs vector retrieval and scoring → Interview/Salary agent drafts questions and outreach → HITL Confirm gate halts before sending.
How is quality ensured?▾
Structured Pydantic schemas, multi-agent cross-checks, confidence thresholds (<0.60 routes to fallback with quick-action options), and full audit logging.
Can we run automation on mobile?▾
Yes. The Expo mobile app mirrors web workflows with offline mutation queuing, push approvals for emails and interview bookings, and analytics dashboards.