AI Agents

Professional services · United States · Agents

1,000 screening calls an hour

720x

screening throughput

from 1,000 calls a month to 1,000 an hour

Client
Silicon Valley technology company, anonymized
Agents
Voice Interviewer
Source
Client call volume before and afterVerified by Proof Ledger, Sep 2026
A recruiter on a video interview at a wooden desk, seen over the shoulder
Screening call · Backend engineerDemo data
Live call · 04:12
  • Thanks, Diego. Tell me about a service you took to production and what broke first.

  • A payments API. The retry logic broke under load.

  • How did you find it, and what did you change?

  • Latency alerts. We added idempotency keys and a backoff.

Rubricquestion 4 of 6

  • Production experiencestrong
  • Debuggingstrong
  • Communicationclear

Written to the ATS

Transcript, score and recommendation · a recruiter reviews the shortlist

VoiceATS
Lines in parallelLine 07 · 02:31Line 12 · 04:12Line 19 · 00:48

Context

The starting point

A Silicon Valley technology company screens every applicant by phone before a recruiter sees the profile. Its team ran about 1,000 calls a month.

Every hiring push outran that capacity, so the queue grew instead of the shortlist.

Screening · before the agentBaseline
Screening calls a month
1,000
Who made them
Recruiters on the phone
Source: Client call volume before the agent

What we built

Built inside the client's systems

01

A voice interviewer agent that runs the structured interview and scores it against the role's rubric.

02

Parallel calling on the DaCodes voice stack, so a month of interviews fits in one hour.

03

Transcript, score and recommendation written to the ATS. A recruiter reviews every shortlist.

Channel

  • Phone call
Voice Interviewer

Runs on: DaCodes voice stackLangGraphBedrock AgentCore

Systems of record

  • ATS
  • Reclutalia
Client cloud
Integrations and channels of this case. Your cloud, your repo, your IP.

Results

What changed

720x

Screening throughput multiplied by 720: from 1,000 calls a month to 1,000 an hour.

Verified by Proof Ledger, Sep 2026

  1. 01

    Recruiters moved from dialing to deciding: they read scored shortlists instead of running first calls.

  2. 02

    Cost and time-to-shortlist enter the ledger when the client signs them off.

Related agents

The agents behind this case

32 agents in the Library

People & talent

Voice Interviewer

Structured voice interviews at scale, scored and written to your ATS. Packaged as Reclutalia.

Verified by Proof Ledger, Sep 2026

Every figure has a row

Exact text, source, date, owner and permission. A row that stops being true comes off this page.

This client asked not to be named. The ledger row identifies it internally.

How the Proof Ledger works
Proof Ledger · row 0002

Published

Claim
1,000 screening calls an hour
Source
Client call volume before and after
Date
Sep 2026
Owner
Practice lead
Client permission
Anonymized
Next review
Oct 2026, AI Council

Which process do we automate first?

Two discovery sessions, no cost, and you leave with two or three quick wins whether or not you build with us.

We reply in under one business day.

Loading the form