The AI stack for a recruiting agency

A recruiting agency's AI stack is about leverage per recruiter: agencies running the full sourcing-and-outreach stack place 2-3x more candidates per recruiter as cost-per-placement collapses. The philosophy is to automate the two operational pipelines — candidate sourcing and hiring-manager outreach — while the recruiter works the warm conversations AI surfaces. The core is enrichment plus message-scoring: AI sources against a job description, scores fit (within EEOC limits), and drafts personalized outreach, with the recruiter owning assessment, references, and the close. Meeting notes that capture intake calls into a structured ATS schema are the second layer, letting recruiters listen instead of type. The defining constraints: EEOC rules mean AI must never filter on protected attributes and screening needs quarterly bias audits, candidate trust is fragile so generic AI outreach backfires on senior roles, and rejection wording (which AI shouldn't send alone) drives reputation more than offers. The mistake agencies make is either avoiding AI (and losing on placement velocity to AI-native competitors) or over-automating the candidate relationship (and damaging their reputation with generic outreach). The stack below targets an agency modernizing its sourcing: enrichment and message-scoring as the core, structured meeting capture for intake, and automation for scheduling — all behind the human judgment that actually places candidates.