meta_pixel
Tapesearch Logo
Log in
Science Quickly

He let AI agents run a start-up—and things got weird fast

Science Quickly

Scientific American

Science

4.41.4K Ratings

🗓️ 6 May 2026

⏱️ 23 minutes

🧾️ Download transcript

Summary

In this episode of Science Quickly, journalist Evan Ratliff joins Kendra Pierre-Louis to discuss his audacious experiment: launching a start‑up staffed entirely by autonomous artificial intelligence agents. Ratliff shares what happened when these agents tried to build a product, manage a human intern, pitch investors and even operate on LinkedIn—sometimes with surprising competence and sometimes with outright fabrication. Recommended Reading: Listen to Evan Ratliff’s podcast Shell Game E-mail us at sciencequickly@sciam.com if you have any questions, comments or ideas for stories we should cover! Discover something new everyday: subscribe to Scientific American and sign up for Today in Science, our daily newsletter. Science Quickly is produced by Rachel Feltman, Fonda Mwangi, Sushmita Pathak and Jeff DelViscio. This episode was hosted by Kendra Pierre-Louis and edited by Alex Sugiura, with fact-checking by Shayna Posses and Aaron Shattuck. Our theme music was composed by Dominic Smith. Learn more about your ad choices. Visit megaphone.fm/adchoices

Transcript

Click on a timestamp to play from that location

0:00.0

Make your summer with Starbucks.

0:04.0

Make it silky, make it creamy, make it ice cool, make it take the scenic route, make it sunset vibes all day, make it soft launch energy, make it smooth, make it strawberry matcha, caramel mocha or brown sugar, make it yours.

0:19.0

Discover the new Starbucks Cloud Frappuccino blended beverage range today.

0:24.3

Their delicious, creamy cold foam will make your summer.

0:27.3

Subject to availability while stocks last. For Scientific American Science quickly, I'm Rachel Feldman.

0:51.8

Have you ever been chatting with a customer service rep and just felt like they're a little off?

0:57.7

Well, customer service is a tough job, so maybe you're the problem.

1:01.5

But it's also possible you were communicating with an AI agent.

1:05.7

These are computer programs designed to autonomously execute tasks.

1:10.2

So while you might use a chatbot powered by a large

1:12.7

language model to answer a specific question using data scraped from the internet, you could give

1:18.4

an agentic AI system a task like, design a website for my new bakery, and expect it to at least try

1:25.1

to accomplish the whole project out in the real world.

1:28.3

Depending on how you design your agent and how much freedom you give it,

1:32.4

one of these computer programs could create its own login on a web hosting service,

1:37.0

scour the internet for examples of good marketing copy about croissants,

1:40.9

generate a few fake photos of kids with too many fingers enjoying cupcakes. You get the

1:45.9

idea. Before you know it, you've got a bakery website, though. Maybe not a very good one.

1:51.2

When global management consulting firm McKinsey and Company surveyed nearly 2,000 people about AI

1:56.3

usage last year, 62% of respondents said their companies were at least experimenting with AI agents.

2:03.3

Now, many of those prospective agents are likely doomed to be faceless customer service reps or

2:08.7

code monkeys. But to hear the AI industry hype machine tell it, agentic AI could replace just

...

Please login to see the full transcript.

Disclaimer: The podcast and artwork embedded on this page are from Scientific American, and are the property of its owner and not affiliated with or endorsed by Tapesearch.

Generated transcripts are the property of Scientific American and are distributed freely under the Fair Use doctrine. Transcripts generated by Tapesearch are not guaranteed to be accurate.

Copyright © Tapesearch 2026.