Call center ai · Production

Cars24 uses ElevenLabs Agents to power India's largest voice-driven car retail operation

The problem

Cars24 lacked visibility into thousands of weekly test drives where mechanical issues, loan objections, and SOP misses went undetected, causing inconsistent customer experiences and hindering regional decision-making. A large contact center also needed to serve predominantly first-time car buyers with timely, high-quality call guidance at scale.

Workflow diagram · grounded in source
1
Test drive audio streamed to Scribe
trigger
“tablets with open-mic recording that stream audio directly to ElevenLabs Scribe for transcription and diarization”
2
LLM pipeline labels field signals
ai_action
“output feeds an LLM pipeline that labels customer, salesperson, car, and hub signals, enabling structured analysis of field conversations”
3
Issue detection and pattern surfacing
output
“detect issues such as mechanical defects during a drive, categorize objections, and surface patterns by SKU, region, or hub”
4
ElevenAgents automates contact calls
ai_action
“Cars24 integrated ElevenLabs ElevenAgents to automate a meaningful share of the contact funnel”
5
Live transcription and TTS response
ai_action
“Scribe Realtime V2 for live transcription - TTS Flash v2.5 for natural, low-latency responses - Workflow orchestration tuned for buying guidance and objection handling”
6
Data-led decisions across regions
output
“Cars24 now drives data-led operational decisions across regions, resulting in measurable business improvements”
Reported outcome

The on-ground intelligence system processed 22,000+ hours of audio, yielding a 35 percent conversion uplift, 20 percent CSAT improvement, and 10 percent hub cost reduction.
The AI voice agent system now handles more than 3 million minutes of calls, achieving 45 percent assisted sales and a 50 percent reduction in calling costs.

Reported metrics
Hours of audio analyzed22,000+
Conversion uplift35 percent
CSAT improvement20 percent
Hub cost reduction10 percent
Show all 9 reported metrics
hours of audio analyzed22,000+
conversion uplift35 percent
CSAT improvement20 percent
hub cost reduction10 percent
AI-supported call minutes3 million+
assisted sales rate45%
calling costs reduction50%
AI call handling target30 percent
first-time car buyers share70 percent
Reported stack
ElevenLabs ScribeElevenAgentsScribe Realtime V2TTS Flash v2.5LLM pipeline
Source
https://elevenlabs.io/blog/cars24
Read source ↗

Frequently asked questions

What did this team achieve with this AI workflow?

The on-ground intelligence system processed 22,000+ hours of audio, yielding a 35 percent conversion uplift, 20 percent CSAT improvement, and 10 percent hub cost reduction.

What tools did this team use?

ElevenLabs Scribe, ElevenAgents, Scribe Realtime V2, TTS Flash v2.5, LLM pipeline.

What results were reported?

Hours of audio analyzed: 22,000+; Conversion uplift: 35 percent; CSAT improvement: 20 percent; Hub cost reduction: 10 percent (source-reported, not independently verified).

How is this call center ai AI workflow structured?

Test drive audio streamed to Scribe → LLM pipeline labels field signals → Issue detection and pattern surfacing → ElevenAgents automates contact calls → Live transcription and TTS response → Data-led decisions across regions.

Related call center ai cases