Agentic recruiting
Recruiting workflows in which an AI system maintains state, uses approved tools, takes bounded multi-step actions, and returns work for recruiter review rather than only answering one prompt.
Recruiting and talent sourcing glossary
Definitions for the concepts behind modern evidence-first sourcing, recruiting AI, talent intelligence, candidate review, source measurement, and recruiter-controlled automation.
Last reviewed: 2026-09-06 · 33 terms · 12 deep reference guides
Candidate 360
A unified recruiter review view that connects identity, work history, role evidence, source records, contact signals, prior recruiting history, and uncertainty without erasing provenance.
Candidate identity resolution
The process of deciding whether multiple source profiles or records belong to the same real person, using deterministic identifiers and reviewable evidence rather than name similarity alone.
Candidate sourcing
The systematic process of translating a hiring need into search hypotheses, finding relevant people across multiple sources, reviewing evidence, and calibrating the search with recruiter and hiring-team feedback.
Contact resolver
A layer that ranks and consolidates candidate contact signals by type, confidence, freshness, corroboration, and prior success or failure instead of dumping every provider result.
Contactability
Whether a contact channel may appropriately be used for recruiting outreach under the source rights, opt-out state, organizational policy, and applicable workflow—not merely whether an email or phone number exists.
Discovery expansion
A synonym, adjacent title, nearby geography, related technology, donor company, or other search concept used to improve recall without changing the literal hiring requirement.
Donor company
An organization likely to produce relevant talent because its work environment, customers, technology, mission, regulation, or operating model resembles the hiring need.
Duplicate pressure
The proportion of discoveries that resolve to candidates already found, reviewed, or known. Rising duplicate pressure can indicate source overlap, query collapse, or genuine market saturation.
Evidence ledger
A structured record of candidate assertions that preserves source, observed time, assertion type, confidence, provenance, and related verification state.
Evidence-backed ranking
Candidate ranking in which the system exposes the evidence supporting each important requirement, plus missing evidence, inference, uncertainty, and verification needs.
Explicit requirement
A hiring criterion directly stated or confirmed in the role brief, such as a required skill, experience floor, location constraint, clearance, credential, or disqualifier.
Field evidence
Observable public, licensed, customer-authorized, or recruiter-confirmed information that supports a candidate claim. Evidence strength depends on source, recency, context, and assertion type.
Hard requirement
A criterion the hiring team has explicitly made non-negotiable. Search expansions and useful adjacent signals must not silently become hard requirements.
Human approval boundary
A point where a consequential AI action—such as outreach, identity confirmation, irreversible data change, or verification-sensitive conclusion—requires explicit authorized human control.
Inference
A claim derived from incomplete evidence rather than directly observed. Inferences can support discovery or ranking but should remain visibly distinct from verified facts.
Market map
An actionable model of the talent market connecting donor companies, title families, geography, adjacent talent, evidence sources, coverage, and next-search opportunities.
MCP (Model Context Protocol)
An open protocol that lets AI assistants connect to approved external tools and data sources through a standardized interface. Recruiting systems can expose search, analytics, and workflow capabilities through MCP.
Normalization
The process of mapping equivalent language to a consistent concept—for example, “RHEL” and “Red Hat Enterprise Linux”—without changing the underlying requirement.
People-data provider
A vendor or API that supplies person, professional-profile, identity, company, or contact data. Provider outputs should retain source rights, freshness, confidence, and usage boundaries.
Provenance
The record of where a piece of data came from, how it was acquired, when it was observed, what kind of assertion it represents, and what the system is allowed to do with it.
Recruiter calibration
The process of converting recruiter or hiring-manager feedback on real candidates into explicit changes to ranking, source priority, exclusions, evidence expectations, or search strategy.
Role Brain
The SourcingOS role-intelligence layer that turns a natural-language hiring brief into structured requirements, normalization, search expansions, geography, verification needs, and other role state.
Search exhaustion
A defensible state in which the major search hypotheses, source lanes, donor pools, geography, title families, and reasonable expansions have been tested and incremental unique yield has materially collapsed.
Search hypothesis
A testable idea about where relevant candidates or evidence may exist, such as an adjacent title family, donor-company cluster, open-source community, registry, or specific source.
Search lane
A distinct sourcing path with its own source, query logic, talent hypothesis, or evidence surface. Lanes should be measurable independently enough to compare contribution and overlap.
Search plan
The recruiter-facing artifact that converts role requirements into title families, donor companies, geography, source lanes, query hypotheses, evidence needs, exclusions, and calibration strategy.
Source diversity
Coverage across meaningfully different discovery and evidence sources rather than superficial use of many tools that all expose the same underlying candidate population.
Source profile
A record representing a person as observed in one source. Source profiles should keep their own identifiers and provenance even when linked to a canonical candidate.
Standing role agent
A persistent sourcing agent that keeps role state over time, skips already-reviewed identities, searches approved lanes, applies recruiter-approved learning, and reports net-new work plus an audit trail.
Talent intelligence
Evidence used to improve hiring decisions about market supply, donor companies, title patterns, geography, skill adjacency, competition, source coverage, or requirement tradeoffs.
Unique contribution rate
The share of useful discoveries from a source or lane that were not already found elsewhere. It helps distinguish incremental sourcing value from raw result volume.
Verification
A separate process for confirming a claim through the authoritative or employer-approved source appropriate to that claim. Public evidence can support discovery without constituting verification.