Standardized hub-incharge screening across six languages, and 83% of the AI’s picks cleared the human interview.
Standardization and authenticity, both at once
A Shadowfax hub incharge runs a last-mile hub: managing riders, parcel sorting, peak-load days, returns and cash-on-delivery reconciliation. These are local roles, hired largely through referrals, which makes two things hard.
Standardization: across thousands of hubs and many regional languages, the same candidate could be judged differently depending on who ran the interview. And authenticity: referral hiring into cash-handling roles is a genuine fraud and misrepresentation risk.
“Quick without quality slipping, at our volumes, the real test.”
One interview, six languages, on the candidate’s phone
Shadowfax put every hub-incharge candidate through one standardized AI video interview, run natively in six regional languages (Hindi, Tamil, Telugu, Kannada, Malayalam, English) on the candidate’s own phone. Every candidate was assessed on the same 12-parameter rubric, and every response was logged and every score explained.
Over 44 days, 634 candidates were screened: 311 (49.1%) cleared the quality bar, 323 (50.9%) were filtered out before HR, and 33 (5.2%) were red-flagged for near-zero responses. Roughly 317 recruiter hours were saved versus manual screening.
hub-incharge candidates screened in 44 days across six regional languages, about 14 a day with no scheduling backlog.
Period: Feb 1–Mar 15, 2026. Source: Shadowfax TA reporting.of AI-passed candidates cleared the human Round 1, versus 1.8% of AI-failed, a 46x difference.
Source: Shadowfax TA reporting.44 days of screening, in numbers
Feb 1–Mar 15, 2026, from Shadowfax TA reporting.From box-ticking to real evaluation
The 12-parameter rubric was tuned to what a hub incharge actually has to do: managing riders, sorting accuracy, peak-load and cash reconciliation. The 83.2% agreement with human interviewers is that calibration proven, not asserted.
The first cut of the rubric was too rigid. Shadowfax initially wanted simple tick-marks, but a checklist did not capture whether a candidate understood the work. The fix moved to a more subjective evaluation, scored consistently.
The agent was set up mobile-first for a field workforce, across six regional languages, with recruiter training. Zeko AI’s auditable screen was paired with Shadowfax’s own verification checklist to surface bad apples before onboarding.
Renewed, and expanded into L&D
Weekly interview volumes held steady (74–118 a week), showing the screen was embedded in the workflow rather than bolted on. Over roughly six months, Shadowfax renewed and expanded into L&D, using Zeko AI to evaluate post-training effectiveness and competency gaps.
As of September 2026 the screen remains live, and the agent now also runs Marathi, Gujarati and Odia beyond the six languages documented here.
Scope: detailed figures cover Feb 1–Mar 15, 2026 (44 days), from Shadowfax’s own Talent Acquisition report to leadership; the broader engagement spans about six months with a renewal and an L&D expansion. “Cleared Round 1” is measured on candidates who reached a decided human-interview outcome. Named and quoted with Shadowfax’s approval.
See what Zeko AI can do for your hiring
Book a 30-minute demo and see how teams like Shadowfax calibrated Zeko AI to their own hiring bar.
Act Now
Build a Consistent, Audit-Ready Hiring Process
Standardize interviews across geographies and improve hiring quality with Zeko's AI platform.
Trusted by 150+ enterprises
SOC2 · GDPR · ISO27001
4.8/5 Average Candidate Rating
Act Now
Build a Consistent, Audit-Ready Hiring Process
Standardize interviews across geographies and improve hiring quality with Zeko's AI platform.
Trusted by 150+ enterprises
SOC2 · GDPR · ISO27001
4.8/5 Average Candidate Rating
Act Now
Build a Consistent, Audit-Ready Hiring Process
Standardize interviews across geographies and improve hiring quality with Zeko's AI platform.
Trusted by 150+ enterprises
SOC2 · GDPR · ISO27001
4.8/5 Average Candidate Rating