Intel / article

How We Scaled Our Customer Interview Process With Elevenlabs Agents

Source: elevenlabs.io
Author: ElevenLabs editorial sitemap
Date: Published Sep 16, 2026

Why it matters

Shows a concrete pattern for voice agents that run research interviews at scale: follow-up prompting, simulation tests before launch, and structured analysis of transcripts. Builders can reuse the approach for their own feedback loops.

Key takeaways · AI-distilled

ElevenLabs ran a voice- interviewer on Gemini 2.5 Flash that held over 230 ElevenReader user interviews in under 24 hours, more than 36 hours of conversation, with 85% of calls judged successful and on-topic.
Each 10-minute interview cost just under $1.00, about 9 cents per minute, according to the post. The median call reached the 10-minute cap enforced by the agent's end_call tool, and the median conversation ran 25 messages.
The pushed for follow-ups on vague or one-word answers, simulated tests checked edge cases such as vague replies and inappropriate language before launch, and the Analysis feature extracted structured answers from each transcript.
Segmenting the results showed 21% of fiction readers asking for multi-character dialogue, well above other segments, and almost 95% of respondents engaged without acknowledging the interviewer was an AI, per ElevenLabs.

Terms in this piece · Glossary

AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
guardrails — The checks around a model that block bad inputs and outputs — filters, validators, and permission rules the model itself can't override.
system prompt — The standing instructions a model receives before any user input — defining its role, rules, tools, and tone for the whole conversation.

Read the source elevenlabs.io