Candidate screening, 1,200 applications, 40 interviews

The challenge

The firm received around 1,200 applications per role. Human screening at that volume was incomplete by necessity, recruiters skimmed, applied criteria inconsistently, and unconscious bias crept in. Strong candidates fell through simply because no one got to them in time.

The sheer manual load meant first contact was slow, and good people accepted offers elsewhere while their application sat unread.

Inconsistency also created legal and quality risk: two similar candidates could get very different treatment depending on who screened them.

The solution

An AI screening agent now runs a structured, consistent first interview with every applicant:

• Structured first interview at scale, the Screening Agent conducts a guided chat interview over Email and WhatsApp with all 1,200 applicants, so everyone gets a fair, complete first pass.
• Scored against real criteria, role criteria held in the Knowledge Base let the agent score consistently, removing the variability of who happened to screen whom.
• Top candidates surfaced with reasoning, the agent shortlists the strongest and explains why, so recruiters focus on a clear, justified top tier.
• Candidates tracked in CRM, each applicant is written to CRM via create_crm_lead with their screening conversation attached via the chat link.
• Fast hand-off to recruiters, shortlisted candidates pass to a human via Hand-off Control with full context, so the recruiter's time goes to interviews, not triage.

From 1,200 applications the firm reliably surfaced its ~40 best for interview, consistently and fast enough that strong candidates no longer slipped away.

What was deployed