For twenty years, sourcing skill meant Boolean skill: nested parentheses, synonym chains, site: operators. That expertise still works — but it's no longer the default, because AI-native search (natural-language queries interpreted by LLMs over profile data) now matches or beats a competent Boolean string on most searches while being accessible to anyone who can describe the role.
Where AI search wins: speed to first credible results (minutes, not an afternoon of string-tuning); conceptual matches Boolean can't express ('engineers who've scaled consumer infra through hypergrowth'); synonym and adjacent-title coverage without manual chains; and accessibility — hiring managers and founders can self-serve. Tools built this way include Juicebox, and attribute-based engines like Findem extend the same idea to career-trajectory signals.
Where Boolean still wins: auditability — you can see exactly why every result matched, which matters for compliance-heavy searches; precision on hard filters (certifications, clearances, licenses) where 'roughly relevant' is worthless; and portability, since a Boolean string works on any platform, while AI search quality is locked to the vendor's model.
The dichotomy is also dissolving from the other side: autonomous agents like Noon make the search-syntax question moot for their users — the agent formulates and iterates queries itself against a calibrated hiring bar, so nobody on the team writes search strings for those roles at all. The skill that appreciates isn't syntax; it's articulating the hiring bar precisely enough that a human or machine can execute against it.
Practical guidance: teams hiring in compliance-heavy or hard-filter domains should keep Boolean skills sharp and prefer tools that expose filter logic (SeekOut is the standard). Everyone else should default to AI-native search or agents and spend the reclaimed hours on candidate conversations — the part of recruiting nothing automates well.