Why your AI agent is upcharging you

SPEAKER_1 (0:00)

This episode is brought to you by Paypal. You know how a mom's bag has everything? Sunscreen, snacks, a stapler?

The new Paypal app is like that, but for your money. Shop, pay, manage your account and earn rewards all in one place. And with purchase protection on eligible items, biometric security and pass keys, you're protected at every step. Download the new Paypal app to get started. See paypal.com/protectionterms.

SPEAKER_2 (0:29)

When you need to build up your team to handle the growing chaos at work, use Indeed Sponsored Jobs. It gives your job post the boost it needs to be seen and helps reach people with the right skills, certifications and more. Spend less time searching and more time actually interviewing candidates who check all your boxes. Listeners of this show will get a $75 sponsored job credit at indeed.com/podcast. That's indeed.com/podcast. Terms and conditions apply. Need a hiring hero? This is a job for Indeed Sponsored Jobs.

SPEAKER_3 (1:00)

When you think about Crocs, you think the classic clock. But did you know there's also a canvas clock, a vegan suede clock, a coastal inspired style, a style with studs, straps and lots of sass.

Honestly, I could keep going with the surprises. You've got to see them for yourself because there is literally a style for every kind of girl. So do your thing and head to crocs.com or a store near you now.

SPEAKER_4 (1:26)

Meta's new AI agent, Muse, hit nearly 2 million downloads in a matter of days, vastly outpacing the early growth of ChatGPT.

We are looking at a piece of software that entirely skipped the usual slow adoption curve where early users just sort of tolerate bugs for a few months. It went straight to mass consumer usage. The raw numbers for Muse basically indicate a shift in what people expect from the hardware they carry around all day. They want a system that executes tasks mechanically in the background.

SPEAKER_5 (1:55)

Yeah, the invisible costs of achieving that kind of immediate traction is staggering though. Mark Zuckerberg built a free virtual machine with memory and compute for every single person who downloads the app. Absorbing tens of billions in reality labs losses just to get to this specific point.

It does make you wonder though, if software is no longer just answering our questions, but actively taking action on our behalf out in the world, who is that software actually loyal to?

SPEAKER_4 (2:20)

That tension is exactly why Pablo Frugo's analysis of Meta's trajectory is so interesting. Especially when you look at the physical hurdles they cleared. He points out a massive stock drop from its high, the failure of Meta AI in Instagram and WhatsApp, and a Sora competitor that just didn't land.

On top of all that, they spent $14 billion for half of scale AI just to bolster their data pipeline.

SPEAKER_5 (2:44)

Well, no hired CEO could have survived that string of expensive public failures. The modern VC trend heavily favors replacing founders at the first sign of turbulence, but Zuckerberg retains unchecked control over the company. He is essentially the last remaining founder king of that era, operating entirely without the fear of being ousted by anxious shareholders.

SPEAKER_4 (3:05)

I do question though if this is a triumph of vision or just the brute force of unlimited capital. They basically perfectly copied a competitor's idea, open claw, and just scaled it for the masses.

They didn't invent the underlying technology, they just had the money required to smooth out the edges and force it into their distribution channels.

SPEAKER_5 (3:26)

Scaling something of this complexity securely and offering it for free is the actual innovation.

I mean, taking a fragile tool and hardening it, open clause brilliant, but it's unstable. And if a website changes a single line of code, it just gets confused and crashes.

Meta had to absorb the cost and technical debt to make those mechanics invisible, which is exactly what drives that unprecedented adoption rate.

SPEAKER_4 (3:50)

You really see that in the frictionless onboarding that Connor Grennan highlighted when he tested Muse.

Muse skips the standard setup interview and simply asks for permission to read your email to learn about you automatically, which is a brilliant bypass of user fatigue. It even includes this small detail of an avatar that visibly types away on the screen while it works in the background.

SPEAKER_5 (4:12)

It is a clever interface trick, but Grennan did have skepticism about long-term adoption due to what we call the discovery problem.

AI adoption is slow because it doesn't serve as a perfect replacement for an existing tool. So users have to actively invent their own use cases. Most people just stare at the blinking cursor and draw blank because they aren't used to managing a digital employee.