A senior recruiter at a global gaming studio opened a single requisition last quarter and got 300 applications in three days. Two hundred of them showed 100 percent match scores in the ATS. None of them were qualified.
This is the AI sourcing paradox. The same models that help recruiters surface candidates also help candidates beat the filters. When everyone has the optimizer, the optimizer stops being a signal.
Traditional sourcing was slow but the friction acted as a filter. A recruiter spent four hours building a Boolean string and combing through profiles. The cost of that effort meant only motivated candidates made it through. AI removed the cost on both sides.
What still works in 2026 is not better filters. It is better signal extraction once the candidate is in the funnel. The recruiter at the gaming studio now uses job titles as the first pass because titles are harder to fake than skill keywords. Then cross-verification against LinkedIn. Then a short structured screen with a human.
The deeper shift is moving the assessment burden left. Instead of trying to evaluate 200 inbound resumes, lean teams now run a 10 minute structured screen and let the interview signal drive the funnel. Tools like Mazle treat every interview as a data point that compounds, which means a 20 person pipeline gives you more usable signal than a 2,000 person resume pile.
Sourcing in 2026 is not about finding candidates. It is about deciding which signal to trust. AI made resumes useless and made structured interviews 10x more valuable.
