Recruiting · Production

Apna scales 7.5 million AI interview minutes using ElevenLabs

The problem

Interview preparation in India was generic, disconnected, and inaccessible to most job seekers. Apna needed to deliver personalized, conversational mock interviews with lifelike timing, empathy, and domain depth at massive scale.

Workflow diagram · grounded in source
1
Candidate speaks
trigger
“Each turn begins with candidate speech”
2
Multilingual ASR and NLU
ai_action
“processed by multilingual ASR and NLU models”
3
Evaluate intent and retrieve data
ai_action
“Workflow logic then evaluates intent, emotion, and context, retrieves most relevant domain data”
4
TTS question playback
output
“composes the next question, and plays it back through ElevenLabs — all within roughly 300 milliseconds end-to-end”
Reported outcome

Apna delivered over 1.5 million AI interviews totaling 7.5 million voice minutes with sub-300 ms latency, democratizing access to high-quality interview preparation for millions of job seekers across India.

Reported metrics
total AI interview voice minutes7.5 million voice minutes
AI interviews deliveredover 1.5 million
End-to-end latencysub-300 ms
TTS playback start latency150–180 ms
Show all 8 reported metrics
total AI interview voice minutes7.5 million voice minutes
AI interviews deliveredover 1.5 million
end-to-end latencysub-300 ms
TTS playback start latency150–180 ms
platform usersover 60 million
companies on platform10,000+
roles covered30,000+
role-company RAG micro-models~500 million
Reported stack
ElevenLabsBlue MachinesRAGASRNLU
Source
https://elevenlabs.io/blog/apna-interview-agents
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

Apna delivered over 1.5 million AI interviews totaling 7.5 million voice minutes with sub-300 ms latency, democratizing access to high-quality interview preparation for millions of job seekers across India.

What tools did this team use?

ElevenLabs, Blue Machines, RAG, ASR, NLU.

What results were reported?

total AI interview voice minutes: 7.5 million voice minutes; AI interviews delivered: over 1.5 million; End-to-end latency: sub-300 ms; TTS playback start latency: 150–180 ms (source-reported, not independently verified).

How is this recruiting AI workflow structured?

Candidate speaks → Multilingual ASR and NLU → Evaluate intent and retrieve data → TTS question playback.

Related recruiting cases