Free Intensive Nov 21–22Free 2-day hands-on Vibecoding Intensive on Nov 21–22 at 16:00 MSK (Sat–Sun) · Live Cohort 3 starts Nov 23Join Free →Cohort 3 (Nov 23)
ReleaseAPI / Product · ActionableElevenLabs Blog→ original

Televisa Foundation Hands 1036 Entrepreneur Interviews to an AI Agent in 4 Weeks

Mexico's Fundación Televisa gave its first-round interviews to ElevenAgents, a voice AI agent, so staff would not have to find time in their calendars. Applicants to the POSiBLE accelerator picked their own time, spent five minutes describing revenue and customers, and got feedback right away. From June 29 to July 24, the agent conducted 1036 interviews, and the team used them to select 70 projects for the final. For screening hundreds of applications, the setup works.

AI-processed from ElevenLabs Blog; edited by Hamidun News
Televisa Foundation Hands 1036 Entrepreneur Interviews to an AI Agent in 4 Weeks
Source: ElevenLabs Blog. Collage: Hamidun News.
Listen to article
Article Fact Sheet & Key Figures
ParameterDetails
Primary sourceElevenLabs Blog ↗
Editor-in-Chief
Key metric #1Using the same agent, Bécalos interviewed 394 young people in two days, July 7–8
Key metric #2The agent is built into the POSiBLE web platform and runs around the clock
Key metric #3Testing began in March, and the connection to the foundation's chat systems went through an API

Imagine having to talk to a thousand candidates, each with a different schedule. That is exactly the problem faced by Mexico's Fundación Televisa, which runs POSiBLE, the country's largest accelerator for entrepreneurs. Instead of juggling hundreds of slots in staff calendars, the foundation gave its first-round interviews to a voice AI agent built on ElevenLabs' ElevenAgents platform.

Applicants opened the agent in a browser at a time that suited them, spent five minutes talking about their business, and received feedback right after the call. From June 29 to July 24, roughly four weeks, the agent conducted 1036 interviews, and the POSiBLE team used those conversations to choose 70 projects for the final stage. If you screen candidates, customers, or program participants, this case shows how to remove the biggest bottleneck: scheduling.

How it worked

Fundación Televisa is a nonprofit with more than 20 years of experience in education, entrepreneurship, and digital inclusion. POSiBLE supports businesses at the early, operating, and growth stages with training, mentoring, and a place in a national community. To keep selection fair, the team needs to discuss growth, customers, and sales with every applicant.

The voice agent was built directly into the POSiBLE web platform. Entrepreneurs described their business, from revenue to customer base, and the average interview lasted about five minutes. The agent is available at any hour, so Andrés Chama, founder of the civic participation platform Comuni, for example, did not have to wait for a staff member to have a free slot. He talked to the agent when it was convenient for him and got feedback immediately.

On the technical side, the foundation's team, led by CTO Julio Alberto Ramos, began testing in March after a demo at the office in Mexico City. The team connected ElevenLabs to the foundation's existing chat systems through an API. In June, the foundation joined the Impact Program, ElevenLabs' program for nonprofits. The ElevenLabs team in Mexico helped configure the agent, estimate how many calls it could handle at once, and integrate it into the platform.

A second project and participant feedback

The foundation applied the same approach to its Bécalos scholarship program. For admissions to the Bécalos English Challenge Movilidad, the agent worked around the clock on July 7 and 8 and interviewed 394 young people in two days.

Participants pointed out two things. Karina Michel, who made it into the cohort, said the agent's speech sounded very natural compared with other voice models. Andrés Chama called the agent's conclusions fairly accurate and the post-interview feedback useful. He also stressed that the agent did not replace the POSiBLE team but let him move at a comfortable pace.

The selected founders met in person at the national POSiBLE camp in Santa Fe (Mexico City) from September 30 to October 4. ElevenLabs staff ran a 90-minute workshop for them on automating business processes with voice AI, and also described the ElevenLabs Startup Grants program for early-stage companies.

What to keep in mind and what to take away

The source is ElevenLabs' own corporate blog, so this is more of a showcase than an independent assessment. The text gives no data on the cost per interview, the time the team saved, the rate of errors and dropped conversations, or how closely the agent's assessments matched human decisions. It is also unknown how exactly the interviews influenced the choice of the 70 projects. Participation in the Impact Program likely comes with special terms for nonprofits, so the pricing from this case cannot be carried over to a commercial project.

The practical takeaways:

  • A voice agent is well suited to short, structured conversations: candidate screening, customer pre-qualification, and collecting initial data.
  • The main gain is accessibility: people choose their own time, and fewer drop out at the scheduling stage.
  • Instant feedback after the call turns the interview from a black box into a useful service for the participant.
  • The integration went through an API into an existing web interface, with no separate app. If you already have a chat system, you can add a voice layer in a similar way.

Before launching, decide who makes the final call. At POSiBLE, selection stayed with the team, which is a sensible setup for decisions with real consequences for people. Compare the agent's scores with the opinions of human interviewers, even on a small sample, before scaling the process to thousands of conversations.

Story

Key Questions Answered

How it worked?

Fundación Televisa is a nonprofit with more than 20 years of experience in education, entrepreneurship, and digital inclusion. POSiBLE supports businesses at the early, operating, and growth stages with training, mentoring, and a place in a national community. To keep selection fair, the team needs to discuss growth, customers, and sales with every applicant. The voice agent was built directly into the POSiBLE web platform.

What are the key details on A second project and participant feedback?

The foundation applied the same approach to its Bécalos scholarship program. For admissions to the Bécalos English Challenge Movilidad, the agent worked around the clock on July 7 and 8 and interviewed 394 young people in two days. Participants pointed out two things. Karina Michel, who made it into the cohort, said the agent's speech sounded very natural compared with other voice models. Andrés Chama called the agent's conclusions fairly accurate and the post-interview feedback useful.

What to keep in mind and what to take away?

The source is ElevenLabs' own corporate blog, so this is more of a showcase than an independent assessment. The text gives no data on the cost per interview, the time the team saved, the rate of errors and dropped conversations, or how closely the agent's assessments matched human decisions. It is also unknown how exactly the interviews influenced the choice of the 70 projects.

ZK
Zhemal Khamidun — CPO at Alpina Digital, product manager of AlpinaGPT. Building AI products since 2019. Hamidun News is a pet project that grew into a production aggregator with 6,000+ readers.
✦ Private AI Builders Community · Vibe Club

Ship AI Faster with Founders, Engineers & Weekly Architecture Reviews

Inside Vibe Club (hamidun.com/club) — weekly live masterminds with Zhemal Khamidun, Tech Radar, full LMS access, and 370+ production Claude Code skills.

Join Vibe Club →Explore Hamidun Academy
What do you think?
Loading comments…