Recruiting · Production

Traba deploys AI interview agents to scale industrial staffing

The problem

Industrial staffing is slowed by qualification requirements, language barriers, regulatory constraints, and variable shift schedules. Traba needed to scale without hiring thousands of recruiters, requiring a consistent, reliable system to assess worker fit faster.

First attempt

Scout V1 was monolingual, used a single LLM for all interview steps, relied on static question sets, produced only a basic one-pass evaluation, and still required human operators to make final decisions.

Workflow diagram · grounded in source
1
Worker triggers interview call
Trigger
Scout conducts phone-based interviews for warehousing, logistics, and manufacturing roles.
▾ source quote
“Scout now conducts over 50,000 interviews monthly across warehousing, logistics, and manufacturing roles”
2
Multi-agent call orchestration
Ai action
Calls are split across specialized agents — introduction, vetting, logistics, and FAQ support — with seamless transitions during the conversation.
▾ source quote
“split calls across specialized agents - introduction, vetting, logistics, and FAQ support - with seamless transitions during the conversation”
3
Dynamic language switching
Ai action
Scout switches between English and Spanish mid-call based on user preference.
▾ source quote
“enabling Scout to switch between English and Spanish mid-call based on user preference. This unlocked access to a previously underserved worker segment.”
4
Question deduplication preprocessing
Ai action
A preprocessing pipeline deduplicates semantically similar questions across interviews.
▾ source quote
“Traba engineered a preprocessing pipeline to deduplicate semantically similar questions across interviews. This reduced redundancy by up to 20% per candidate.”
5
Custom evaluation framework
Ai action
Custom Scout defines what 'good' answers look like on a per-question basis.
▾ source quote
“Traba built Custom Scout, a framework to define what 'good' answers look like on a per-question basis. Evaluations now align with each client's unique criteria.”
6
Ground truth feedback loop
Feedback loop
Human-verified datasets generated through Langfuse enable A/B testing of prompts against real-world performance for fast iteration at scale.
▾ source quote
“By generating human-verified datasets through Langfuse, the team could A/B test prompts against real-world performance — allowing fast iteration at scale”
7
Automated vetting output
Output
85% of all worker vetting across the platform is fully automated.
▾ source quote
“85% of all worker vetting across the platform is fully automated”
Reported outcome

Scout now conducts over 50,000 interviews per month with 85% of worker vetting fully automated, saving over 4,000 operator hours per month, and AI-qualified workers show 15% higher shift completion rates than human-qualified workers.

Reported metrics
Monthly interviews conductedover 50,000
Worker vetting automated85%
Operator hours saved per monthover 4k operator hours per month
shift completion rate for AI-qualified vs human-qualified workers15% higher
Show all 7 reported metrics
monthly interviews conductedover 50,000
worker vetting automated85%
operator hours saved per monthover 4k operator hours per month
shift completion rate for AI-qualified vs human-qualified workers15% higher
interviews completed by March 2025over 17,000
manual vetting hours saved by March 2025more than 1,400 hours
question redundancy reduction per candidateup to 20%
Reported stack
ElevenLabsScoutLangfuse
◆ Does this fit your context?

Compare to your context

Tell us your scale, team, and constraints. We'll show what changes at your size, what fails at your scale, and whether this case is a fit, needs adaptation, or won't scale to you. Free demo, no signup.

Compare to your context →
~30 seconds · free
Source
https://elevenlabs.io/blog/traba
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Scout now conducts over 50,000 interviews per month with 85% of worker vetting fully automated, saving over 4,000 operator hours per month, and AI-qualified workers show 15% higher shift completion rates than human-qu…

What tools did this team use?

ElevenLabs, Scout, Langfuse.

What results were reported?

Monthly interviews conducted: over 50,000; Worker vetting automated: 85%; Operator hours saved per month: over 4k operator hours per month; shift completion rate for AI-qualified vs human-qualified workers: 15% higher (source-reported, not independently verified).

What failed first in this deployment?

Scout V1 was monolingual, used a single LLM for all interview steps, relied on static question sets, produced only a basic one-pass evaluation, and still required human operators to make final decisions.

How is this recruiting AI workflow structured?

Worker triggers interview call → Multi-agent call orchestration → Dynamic language switching → Question deduplication preprocessing → Custom evaluation framework → Ground truth feedback loop → Automated vetting output.

Related recruiting cases

WHAT TO DO WITH THIS

Now compare it to your context

This case is one data point. Whether its pattern fits you depends on your volumes, your stack, and your exception load — that comparison is the step no case study can do for you.

Browse the Recruiting cases →