High Signal Podcasts Evidence ledger
Method
Browse
← Back to evidence

Evidence receipt / preference

Published · transcript-backed

William Beauchamp: preference

26 Jan 2025 Latent Space Outlasting Noam Shazeer, crowdsourcing Chai AI with >1.4m DAU, and becoming the "Western DeepSeek" — with William Beauchamp, Chai Research

“Exactly, exactly. So I think if you want to make audio work, it has to be a unique, compelling, exciting experience that they can't have anywhere else.”

— William Beauchamp

Source trail

Everything needed to verify it.

Speaker
William Beauchamp
Attribution
Verified speaker
Claim type
preference
Recorded
26 Jan 2025
Publisher
Latent Space

Transcript context

…insanely good at execution. So you were first with voice? We were first. We were first. I only know when character launched voice. They launched it, I think they launched it at least nine months after us. Okay. Okay. But the team worked so hard for it. At the time we did it, latency is a huge problem. Cost is a huge problem. Getting the right quality of the voice is a huge problem. Right? Then there's this user interface and getting the right user experience. Because you don't just want it to start blurting out. Right? You want to kind of activate it. But then you don't have to keep pressing a button every single time. There's a lot that goes into getting a really smooth audio experience. So we went ahead, we invested the three months, we built it all. And then when we did the A-B test, there was like, no change in any of the numbers. And I was like, this can't be right, there must be a bug. And we spent like a week just checking everything, checking again, checking again. And it was like, the users just did not care. And it was something like only 10 or 15% of users even click the button to like, they wanted to engage the audio. And they would only use it for 10 or 15% of the time. So if you do the math, if it's just like something that one in seven people use it for one seventh of their time. You've changed like 2% of the experience. So even if that that 2% of the time is like insanely good, it doesn't translate much when you look at the retention, when you look at the engagement, and when you look at the monetization rates. So audio did not have a big impact. I'm pretty big on audio. But yeah, I like it too. But it's, you know, so a lot of the stuff which I do, I'm a big, you can have a theory. And you resist. Yeah. Exactly, exactly. So I think if you want to make audio work, it has to be a unique, compelling, exciting experience that they can't have anywhere else. It could be your models, which just weren't good enough.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.

Search evidence