Episode Details

Retrouvez l'épisode en version française ici : https://www.gdiy.fr/podcast/mati-staniszewski-vf/

ElevenLabs is an iceberg.

People think it’s a startup that uses AI to generate voices, but that’s actually just the tip of the iceberg.

Voice agents that can recreate the voices of throat cancer patients so they can say their wedding vows, identify signs of Parkinson’s in your voice years before a diagnosis, and above all...

a staggeringly complex system of gears, invisible to the user.

The thesis of Mati, its founder, is simple: writing is not the natural language of humans.

Keyboards, screens, and the interfaces we’ve been using for 30 years are historical anomalies on the verge of disappearing.

Just four years after its founding, ElevenLabs is already valued at over $11 billion, with $600 million in ARR and a team of more than 600 people.

A European company that wants to remain European in a sector dominated by Americans.

In this episode, we pull out all the stops:

• Why voice will become the primary interface between humans and machines
• What ElevenLabs really does: agents, healthcare, education, IP, data sovereignty
• Why Mati has placed an engineer on every non-technical team
• The role of humans in a world where 80% of code is already written by machines

And even if this hasn’t piqued your curiosity, you MUST listen to it.
It’s essential for understanding where we’re headed and, above all, how to take action right now before it’s too late…

You can contact Mati on LinkedIn, X, and Instagram.

TIMELINE:

• 00:00:00 : Mati’s first lines of code and the early days of ElevenLabs
• 00:10:57 : His previous experiences at BlackRock and Palantir
• 00:18:45 : Leaving Poland at age 18 and discovering his ambitions
• 00:26:35 : How AI that speaks works
• 00:35:42 : Giving a voice back to people who have lost theirs
• 00:40:01 : Breaking down language barriers
• 00:48:28 : Humans aren’t meant to write, but to speak
• 00:54:03 : The end of screens?
• 00:59:35 : Fused or cascaded agents: the real trade-off
• 01:15:30 : What BlackRock and Palantir taught him—and what’s changing in 2026
• 01:24:37 : The platform, enterprise clients, and the business model
• 01:32:15 : The role of humans in a world of AI agents
• 01:39:31 : How to put AI to work in education
• 01:47:43 : Darth Vader, Epic Games, and growing pains
• 01:53:29 : Passing the Turing test and staying European

We referred to previous GDIY episodes :

• #553 - Paul Frambot - Morpho - Briser le cartel bancaire avec 600 lignes de code
• #543 - Yann Le Cun - AMI Labs - Rendre l'IA plus humaine
• #507 - Laurent Alexandre - Auteur - Vers la fin des études supérieures ?
• #397 - Yann Le Cun - Chief AI Scientist chez Meta - L'Intelligence Artificielle Générale ne viendra pas d