$20 Billion OpenAI: Voice Erases Screen Dependency  ## Document Notes ## Highlights - **OpenAI Is Betting On Audio-First Future** - Jaeden Schafer highlights OpenAI rebuilding audio models from the ground up to power upcoming audio-first devices. - This signals a broader industry shift away from screens toward voice-driven interfaces across contexts. Transcript: Jaeden Schafer There's a new report from the information that says that the company has spent the last two months consolidating engineering product research teams to rebuild its AI audio models. And they're doing this from the ground up. This is all ahead of an audio first personal device, which we're expecting to come out in roughly a year from now. There is a lot going on in the story and a lot is at stake. So we're gonna be covering all of that on the podcast today. If you want to try the latest AI audio models from OpenAI or from 11 Labs, which in my opinion is the best at the moment, I would recommend going and checking out AIbox.ai. Have a playground that lets you access over 40 of the top models, everything from OpenAI, Meta, Google, DeepSeek, tons more. All these audio models, they're on there, and you can try them out for $20 a month on my own startup, AIbox.ai. I'll leave a link in the description if you want to go try that out. This big focus that we're seeing from OpenAI right now to push audio AI, I think shifts or essentially is kind of showing what's happening in the broader tech world. ([Time 0:00:10](https://share.snipd.com/snip/155f710f-dc80-4768-8655-6ed6d0141a79)) - **Screens Fade As Audio Gains Momentum** - Jaeden argues screens are fading as designers and VCs push against screen addiction and favor audio interactions. - Big players like Google, Meta, and Tesla are embedding voice features into glasses, search, and cars, reinforcing the trend. Transcript: Jaeden Schafer Right now, screens are starting to fade into the background. They're not, you know, the trendiest thing. Most of these, a lot of these Silicon Valley startups and venture capitalists, a lot of people are talking about how screen time or screen addiction is bad. And so it feels like people are kind of pushing away from screens. And with that, audio is kind of taking the forefront. It feels like this is the, you know, the response. There's, of course, smart speakers that have already normalized voice assistants. And those are in over a third of U.S. Households. Meta recently added, of course, a feature to their Ray-Ban smart glasses. They use a five microphone array that essentially isolates a voice in a noisy environment. So if you're talking to someone in a bar or at, you know, in some sort of noisy store or market and you look at them, it has these five microphones, it isolates their voice, and in your ear Will amplify just their voice. So effectively, it is, you know, turning your head into a directional microphone. Google also started testing audio overviews in June, which is, you know, covering the search results into conversational summaries. ([Time 0:01:06](https://share.snipd.com/snip/ebbabe00-d657-4ddf-b4bb-fef08a949ba5)) - **Examples Of Screenless Wearable Experiments** - Jaeden recounts several startups building screenless wearables like Humane's AI Pin and Friend's pendant that recorded life audio. - He mentions AI-powered rings from Pebble's founder and Soundbar as further examples of this hardware wave. Transcript: Jaeden Schafer There is a collection of startups that are chasing the same idea. There's, you know, some of them that have, you know, been successful and others that have, it feels like failed extraordinarily, including Humane's AI pin, which burned through hundreds Of millions of dollars before essentially becoming a bit of a cautionary tale for screenless wearables. There's of course the Friend AI pendant. This is a necklace that you wear around your neck. It got famous because the CEO bought friend.com for like $6 million and basically blew their entire fundraising round on the domain name. And with that, their necklace is essentially just sitting there and it records your life and it keeps you company. And, you know, some people are curious about it. A lot of people are worried about privacy concerns. You know, it's a crazy product. So I think right now there's at least two more companies, including Sandbar and another that's led by the Pebble founder, Eric, which if anyone remembers Pebble, this was like the biggest Kickstarter of all time. Back in the day, I think there was like $10 million for a smartwatches as before the Apple watch came out. Very cool, novel idea. And they are developing an AI powered rings. ([Time 0:02:22](https://share.snipd.com/snip/0ff905f8-fd1f-4690-953f-6ea7fcdde1af)) - **Designing Companions, Not Just Utilities** - OpenAI aims to create audio devices that act less like utilities and more like companions, influenced by Johnny Ive's design philosophy. - The company acquired Ive's firm and plans device lineups intended to reduce device addiction via screenlessness. Transcript: Jaeden Schafer So right now, I think this is why OpenAI is putting a massive focus on an upcoming audio model. It's expected early this year, so 2026. And it's reportedly designed to sound more human, handle interruptions like a real conversation partner, and even speak over you mid-sentence, which is kind of interesting slash Annoying. I'm hoping you can turn that feature off if possible. I know like the goal here, they're like, well, a human wouldn't like let you just keep talking forever and ever. But it's like, well, sometimes I would like to, like recently I was using the chat feature on ChatGPT today. And I was like, okay, I want you to do XYZ thing. And I was like, okay, I can do that thing. And I like just kind of cut it off because I'm like, great, whatever. Here's everything you need to know. And I'm like, just listening, just tons and tons of information for this specific problem, trying to give it all the context, kind of the same way I would copy and paste like a massive Document in. And I can only imagine if like halfway through that, it was interrupting me to ask like follow-on questions where I'm like, no, just listen to all my information first, wait till I stop Speaking. So I don't know. They're trying to make it sound more human, but you can imagine ways that that will also be more annoying when you're trying to use it to get some work done. Anyways, we'll see how customizable that is. They said that they are also going to be imagining an entire lineup of devices, including glasses or a screenless speaker that behave less like utilities and more like companions. ([Time 0:04:07](https://share.snipd.com/snip/17038db7-e186-4870-bd3b-8c8e0d17875d)) - **Humanlike Audio With Interruptions** - The expected OpenAI audio model aims to sound more human, handle interruptions, and even speak over users mid-sentence. - Jaeden notes these human-like behaviors could be helpful or annoying depending on customization options. Transcript: Jaeden Schafer It's expected early this year, so 2026. And it's reportedly designed to sound more human, handle interruptions like a real conversation partner, and even speak over you mid-sentence, which is kind of interesting slash Annoying. I'm hoping you can turn that feature off if possible. I know like the goal here, they're like, well, a human wouldn't like let you just keep talking forever and ever. But it's like, well, sometimes I would like to, like recently I was using the chat feature on ChatGPT today. And I was like, okay, I want you to do XYZ thing. And I was like, okay, I can do that thing. And I like just kind of cut it off because I'm like, great, whatever. Here's everything you need to know. And I'm like, just listening, just tons and tons of information for this specific problem, trying to give it all the context, kind of the same way I would copy and paste like a massive Document in. And I can only imagine if like halfway through that, it was interrupting me to ask like follow-on questions where I'm like, no, just listen to all my information first, wait till I stop Speaking. ([Time 0:04:13](https://share.snipd.com/snip/4b70dd8b-6005-4088-95f7-9fc2eb1b8a2d)) - **Use Voice AI For Utility, Not Replacement** - Use voice AI as a productivity tool for tasks like business coaching rather than replacing human relationships. - Leverage conversational models to analyze data and get non-biased feedback on priorities and revenue streams. Transcript: Jaeden Schafer There's a lot of things at stake here. I'd be very curious to see if the if the companionship one really is that popular. Like I definitely can see value in using these for utility. But companionship is interesting. I will say recently, I had a very, very long conversation. It was, of course, the new year. And so I had a very long conversation with ChatGPT where I basically, and this wasn't chat, but parts of this, I used the voice mode to give a bunch of details, but I basically went in there And gave it every single revenue stream I had last year and how much money I made from every single one and gave it a bunch of business ideas I had for this year and asked it where I should Focus my time, what, you know, opportunities I was leaving off on the table, basically asking it for like, sort of like financial advice to break down all my businesses. It did an incredible job told me areas where I needed to focus more should focus less on. Like, I think intuitively, some of those things I like, I could have told myself that but it's so hard when you have these ideas that you feel sort of like emotionally attached to. And it feels good to have a third party that feels non-biased, just listen to the data and give you a give you an output. So like, when we're talking about having a companion, and when Johnny Ive talks about having a companion, like, yeah, if my like, I mean, I use chat GPT for that. But like, if I had some sort of, you know, speakerless device on my desk, where I was just talking to it, and it was responding, and it was like, business coaching me, very useful. ([Time 0:06:50](https://share.snipd.com/snip/932dde51-afb9-44b2-8ac7-2b9672a848ff)) - **11 Labs Is Strong But OpenAI Could Steal Users** - Jaeden endorses 11 Labs as a leading audio provider and shares his experience using its high-tier plan for audio generation. - He remains open to switching if OpenAI produces a competitive audio product. Transcript: Jaeden Schafer And you know, obviously, we're humans, we need humanity, we need people around us. So anyways, off my philosophical soap box there for a second. This is going to be interesting to see what OpenAI comes out with. I'm super excited for just better audio models. Obviously, I think 11 Labs is doing a phenomenal job. And 11 Labs is doing some really interesting things here where they're allowing you to clone your own voice. And they have a ton of really cool tools and features that are like really specifically built for audio. So that's where I find myself going to. I think last month I was on 11 labs, $1,300 a month, uh, AI audio tier. Cause I was getting a bunch of virtual assistants to generate a bunch of audio for a project I'm working on. ([Time 0:08:56](https://share.snipd.com/snip/adfa06e0-e3e0-4e1f-b6b0-cee6105e599c))