Why Sandbar thinks it’s voice-enabled ring can avoid the AI hardware graveyard

0:00This episode of Equity is brought to you by Schneider Electric. Schneider Electric, your energy technology partner. Hello and welcome back to Equity TechCrunch's flagship podcast about the business of startups. I'm Rebecca Balan and this is the episode where we bring on industry experts to help us explore a trend in the tech world and dive deep. AI note-taking hardware has taken off over the past year or so. We've got credit cards, size devices, pendants, pins, even transcribing earbuds, all of which have shown up promising to capture your meetings and turn them into summaries and action items.

0:31A whole wave of wearables, rings especially, are betting that people want to capture stray thoughts and ideas the same way. Today's guest is one of the founders building in that space, and he's got the funding and early traction to back it up. We're joined by Mina Fahmy, co-founder and CEO of Sandbar, the startup behind the stream ring. He's going to do some live demos for us today, so if you're listening, you can also watch on YouTube.

0:58Mina, welcome to the show. Thanks so much for having me. Yeah, I'm really excited to talk about this. So for those who don't know, because why would you? Mina and I met last week at a Betaworks VC event where there was a bunch of different hardware startups. And I, you know, got to see Mina using the Streamring and thought it was really interesting. And I just wanted to, you know, bring you on to talk about werewolves in general and your startup in particular. And before we get into that, though, why don't you give us a little bit of your background? Like, how did you come to starting Sandbar?

1:30Totally. So I grew up in farmland in Maryland. Didn't have a whole lot to do, and so I consumed a lot of sci-fi. And through that, fell in love with this idea of what I later learned is called human-computer interaction. So how do humans and machines relate? And I got my start in neural interfaces, doing research at MIT in the MIT Media Lab, and then moved to New York to join a startup called Control Labs that was developing wrist-worn devices that used deep learning models to translate your gestures into inputs. And you could do really magical things like swiping your thumb to swipe through a UI.

2:13You could write with your fingertip. And later, Meta acquired that startup, and that technology went on to become the neural band that ships with the Meta displays. While at that startup and at Meta, I also met my now co-founder, Kirak. And in parallel with all of this, I saw what was happening with large language models and thought, this is the future of HCI. This is the future of interaction. But as I started to use them, I realized that had perhaps been a bit of an ambitious thought because practically they were very hard to use. Walking around living in New York, I would sometimes want to talk through ideas using these tools or use them to help me remember things, but there was no effortless way to do so.

3:00I didn't want to pull out my phone and unlock an app every time I had a thought. I didn't want to shout in public into my earbuds. I think on a more personal level, I also cared a lot about how I related to this machine because the apps that I was trying largely acted like companions or co-workers or replacements of people in my life. And I was excited about human-computer interaction because I believed it was a way to augment us, not to replace us. And after speaking with Karak about these themes, we thought, let's try to explore what a different relationship could look like. Yeah. So what does it look like?

3:40I mean, you mentioned some of the gestures and things like that. I've seen you do that, right, when you were using the Ring, and you'll give us a little bit of a demo in a bit, but how much of your approach with Stream is a direct evolution of the work you did at like Meta slash Control Labs, or Control Labs, which was acquired by Meta in 2019? I think there were a few core design beliefs. One is the idea of human input and human intention being very important. In the market, you see a lot of devices and software that just records everything. But the downside of that is you don't know what the person actually thinks or wants.

4:23You have to make assumptions. And when you look at computers historically from the command line interface that mainframes were operated by, or the graphical user interface that we're talking through on our computers, or multi-touch on your phone, they were all about human-driven input. Your keyboard, your mouse, your touchscreen. screen. So I think you being in control was very important to us. The thing that was perhaps more unique to Sandbar was this belief we had in what we call self-extension, which is the relationship that we see as being very different from companionship or a coworker. The question basically is, what does it take for a piece of software or for a device to feel like it's a part of you.

5:11And when we started, we tried building a lot of devices, a lot of apps, but we were always asking ourselves the question of how can this feel like it is a more effortless part of you. Okay. Yeah. So why don't we actually just go right into a bit of a demo because I want to, I want, you know, our audience to kind of really understand how it does feel like more of a part of you. Totally. So after a lot of searching, we've now arrived at the first product we're shipping, which is called Stream. And Stream is a private voice ring. So it has a flat touchpad on the side, a mic on top, and a haptic that vibrates when you interact.

5:53And you just hold the touchpad, speak closely into it, and release. And there's three primary use cases of Stream, capturing personal notes, chatting in the moment, and dictating across apps. So I guess a really quick version of that is maybe if I wanted to remember an idea for later or a to-do, I could simply say, add sparkling water to my grocery list. And then as that input goes in, I feel it vibrate three times, which means stream is doing something for me. so that if I open the stream app, I can see it captured what I said. It found my grocery list. And when I tap into it, I can see it added sparkling water right under some dates and hot honey.

6:46Yeah, essentials. So, okay, so then where is this all coming to? So one of my problems with like note-taking apps is that often they all, you know, they'll live in a place that's not necessarily convenient for me, right? or maybe it doesn't translate across my computer and my phone and I can't access it from everywhere. So where does, like when you send any kind of note or reminder or what have you, where does it live? So right now it lives within the stream app. And then what's really important for us is making it easier and easier to take that data wherever you want to go. We think of ourselves as trying to build bridges, not destinations.

7:21So today you can find it whether in your main stream of information or if you swipe over. Let me make sure I don't have any totally compromising information on my screen. I might not actually show the artifacts. This is my personal account. But there's a dedicated notes view in the app. You can also access it on the web. and we're also working on more developer tools and other user-friendly tools to export that data to other places where you might save your notes or information. Yeah, I think when we were chatting the other day, you showed me that it could show up on, there's a web app or... Yes.

8:03Yes, so like on Chrome, you can go to the website and then you can log into your specific profile and access everything, right? Exactly. One thing we see with beta users is some of them will, on their walking commute or on their driving commute, they'll be talking with Stream about their day or about a proposal or an email, and they'll summarize those threads into notes. And then when they get to their computer, they'll pull up the web app and they'll copy it into Cloud or into Notion or into their email client. So it can basically serve as a jumping off point for all of the work or thoughts they had on their commute.

8:43Do you imagine some kind of connection directly into Cloud or Slack or something like that? Absolutely. I think we really want to make it very easy to plug and play. Step one was just getting it working. And then that's what we want to work on right after. Another thing that I thought was cool about it was that you could be, you know, first of all, the whisper aspect was really interesting. You were really, I couldn't hear a word you were saying and you were whispering right into it and it was showing up on the app. But then another thing was how it translated from your phone to the computer. So if you put your phone down, you open up your computer, you wanted to go to Chrome, you could whisper what you wanted to Google and it would show up there.

9:23Like, how do you get that kind of integration? Yeah, so we built those features actually in particular for the other two use cases of Stream, which are chat in real time and then dictation across apps. with chat even if your phone is in your pocket and earbuds are in or it's connected to your car speaker you can actually talk through ideas in real time so if i were to say to stream um i'm trying to remember how i usually describe inner voice um i'm actually in a podcast right now uh can you uh remind me of that

10:04from the fashion tech panel prep back in june you described it this way stream's voice mirrors your own to collapse the distance between thinking and being heard it's designed to disappear and the deeper framing from around that time i can tap to cut it off and we developed a custom whisper model that you saw because we found that when you're chatting privacy is really important so that you can actually tap into the most real ideas that you're having. So even if right now I'll be like right next to the mic.

10:35He's whispering. And then. Ha, and yet here we are coming through fine. Yeah, okay, so this is not, are you not using WhisperFlow or something? Like what's the tech behind the tech? So we use different models on the back end for a combination of transcription directly to stream. And then the third pillar, which is dictation, which is the ability to type into any app. And that's a mix of partner models and models that we've developed in-house. But we're not connected to Whisperflow or that. I can show you, for example, here, this is just my in iOS keyboard. and in dictation I can tap, speak, and it will just immediately go through.

11:23And the really nice thing here on iOS is you don't have to do app switching, which is something all other software-based dictation apps require where you have to swipe back and forth really quickly. It also lets you free up your hands and speak a lot more quietly, which is really nice on the go. This has basically replaced my phone's keyboard. And then the other thing that you saw, which is a little bit harder to demo over a video call, is if I put my phone down and then now I'm on my laptop, I can immediately start dictating and my words will just flow into my laptop. So it has this nice cross-platform aspect that makes dictation really fast.

12:04Yeah, it seems like you're thinking a lot about user experience. Like it's really, what do you think if you're just taking a step back, looking at what you're doing and looking at what others are doing in the space, what is the killer app that will make someone change their hardware ways, right? It's hard to adopt new hardware. Not everyone wants to do that. It really needs to be seamless. So what is it that you think really makes a difference? What moves our world forward? Energy. History has been shaped by it, revolutions won with it, and progress dependent on it. For 190 years, Schneider Electric has been a leading force in this story, creating the technologies that harness energy and transform the competitive landscape.

12:43And today, in the dawning era of energy and industrial intelligence, Schneider is helping customers build more efficient, resilient, and sustainable operations across every sector. From data centers to homes, from buildings to industry. Schneider Electric, your energy technology partner. Discover more at se.com. I think about it in at least two ways. One is we're trying to solve a problem that can't currently be solved with existing software or hardware. And the simplest way I would describe this problem is most products today cannot keep up with the speed of thought, especially for people who are very on-the-go lives or who are verbal processors.

13:33So, for example, a lot of our beta testers and early customers are working moms who are juggling a ton of responsibilities throughout their day. And Stream is really designed to be used in motion when you're dancing between things. We've also seen that, for example, for people with dyslexia, they're some of our most active beta users because most of the tools for thinking that people have are based around writing and reading. And so one user in particular described this as almost unlocking his ability to express himself in a way he couldn't before. I think the other way I think about it is let's look at what is succeeding in software that could be made all the more better with additional hardware.

14:18The primary voice use cases are generally voice notes, voice chat, and voice dictation. What if we just made all of those way better and way more accessible? In the same way that multi-touch made the browser accessible on the phone in mobile contexts, whereas before the browser could only be used in a seated at home context. That's where I think hardware really makes a new experience. Okay, what's the final form factor? Is this always going to be a ring? It's a good question. We lovingly call the ring the mouse for voice because we think that it does for voice what the mouse did for the graphical user interface.

15:02But I'm not sure that it's necessarily the only form factor. Like from a human computer interaction lens, I think a lot about what are the inputs and outputs of information that might be useful to a person. And sometimes you need to see something. You know, I don't think anyone enjoys audio navigation the same way they can just look at a map and know where they need to go. So we're really excited about rings. We think there is a pretty enormous amount of opportunity within them. But historically, we've always worked on and continue to work on other devices beyond them. Okay. And then you might have already answered this, but I guess I'm still thinking about it.

15:40Like, what is it that will get this out of the tech enthusiast world and into the mainstream? Like, Sandbar is a New York-based company, right? You're not in the thick of it in Silicon Valley. So, you know, your customers might be a little bit different. I mean, you mentioned some working moms. That's pretty cool. They could be tech enthusiasts too, though. But yeah, what do you need to do to get this to a point where, like, mainstream people want to use it? It's almost like, why else would we do this? I think that's a really big reason why we're in New York. It allows us to really like focus and be free from a lot of the other pressures being in the Valley create.

16:18I think one thing that's quite nice is we're definitely seeing a lot of excitement around voice in tech. But in many ways, it is a more generally useful tool than interfaces that came before it. Like if you think about the amount of training you need to even master QWERTY on a keyboard versus simply speaking, which we all gained as we grow. I actually think this will be mass market much faster than many expect. When we announced, we got a lot of excitement from within tech. But since then, our customers and increasingly the people in our beta program are majority outside of tech. and that has been really intentional for us.

17:09Yeah, I guess, you know, thinking about this question a different way, you know, there's been so many voice-based hardware devices that have emerged aimed at like companionship, productivity, personal growth. You know, there's the card-shaped devices like Plot or Pocket. You had Friend or Limitless and Taya. Those were more like high-end jewelry pendants. Omi, that can be worn as a necklace or taped to your head as a brain interface. There's a wristband from Bee, which is now part of Amazon. So there's a lot going on. Pebble, I think, is probably one of your more direct competition. But, you know, a lot of these companies aren't doing too hot.

17:45Some of them really failed, like Humane sold to HB. You know, Rabbit was trying to fix some things. VS Software Update. Friend had some serious backlash. So, yeah, how are you thinking about things differently? Or where do you think they went wrong? Yeah. So I was just smiling because I really appreciate how much you're in it. I'm a bit of a nerd for this industry, but I really appreciate it. The thing that I mentioned that is really important for us is keeping the human in the loop. And I think for a lot of these devices, they really weren't designed for human interaction. They were more passive devices.

18:28and some of them are doing incredibly well for their use case. It's just not clear to me whether use cases like meeting recording will be mass market and ubiquitous, especially outside of very local work settings. Our head of marketing was telling me that she was at, I think, a grocery store and a lawyer saw her speaking into the ring and said, oh, is that Stream? And she was, you know, really had. Like, wow, like, you know, no one knows about us. Like, that's so cool that you recognized it. And he said, well, yeah, I actually, I'm not going to name them, but I have one of these meeting recorders given to me by my company.

19:08And it's useful and all, but I wouldn't wear it outside. Whereas like this, I like really want. And that I think is really motivating and encouraging for us because we are trying to make something for general use rather than for the very specific business setting. I think related to whether the person actually interacts with it versus it being a passive device is also whether it makes you feel beautiful and look good or cool as you're using it and as you're wearing it. That's, for example, why stream is pushed to talk rather than continuously recording, because we know that people will be much more comfortable if that's the case.

19:46And that social acceptability is perhaps the most important attribute of a wearable. Yeah. What do you think about the always on thing that's going on? Like there's quite a few companies that are promising to be always on, whether it's, you know, a set of glasses or a ring or a wearable or note taker. And it's like, that feels weird. Like, I don't I don't know who wants that. But what are your thoughts there? For something that's truly always on, I don't know. The thinking kind of escapes me. Um, uh, I sometimes we'll hear people working on that say social conventions will change, but then I'm kind of met with the question of like, why would they, who wants this?

20:30Yeah. Um, so I don't know, I guess I, I'm a firm believer that we get to choose the future we want to be part of. And I enjoy these kinds of conversations in part because it allows us to question whether a always recording future or a privacy optional future is predetermined or something that we get to opt into or out of. Say more about that because there's so many more mics around us right now. And I think it's interesting that yours, it's not going to pick up my voice when it's recording your voice, right? Yeah, it's really designed to be close. Yeah. Yeah, there's just so many more microphones around us, people constantly picking up information like how do you deal with data and privacy?

21:12For this, it's actually, it really comes from that North Star of self-extension. So the question we keep asking ourselves is, what would something need to be like in order for it to feel like a part of you? Push to talk is part of that. Push to talk is what allows it to feel like it's under your control. The question I asked Stream before about inner voice is because Stream's default voice is partially personalized to yours. So it feels a bit more natural as we found testing it on ourselves and on others. And that extends to data as well. So, you know, we do a lot of industry best practices, like your data being encrypted in transit and at rest.

21:48We have zero retention with any, you know, processors that we use. And in general, we're trying to just march ever closer to severing us as much as possible from the data so that it can be as private to you as it can be. Yeah, interesting. I'm not sure if you have a comment on this at all. But, you know, Earlier this week, Mark Zuckerberg published a letter about personal superintelligence kind of laying out his vision. This is in line with Meta releasing Muse Glimmer. So that's their open, essentially an open version of their very powerful closed model, Muse Spark, that they released in April.

22:28and you know it's it's kind of a glimpse into this vision of of super intelligence run locally on your own device an extension of yourself always their companion or not he doesn't call it a companion he calls it a capable agent but i guess i'm just kind of wondering like how you see wearables fitting into do you agree with that kind of future and do you see wearables fitting into it a lot of this is actually very practical for us like when you interact with stream it's running many models in parallel. It's processing your input. It's providing a response. It's forming memories that it's retrieving.

23:05It's taking actions like creating notes or setting reminders. It's doing transcription. So in practice, we use many open and closed source models because they have different strengths and weaknesses. And I expect that to continue into the future. I think it's very unlikely that we hit a future where it's just closed source models or it's just open source models. There will always be more compute available to closed source models. And at the same time, there will always be certain benefits potentially around cost or around speed for models where you can more directly control where they're hosted, which is in the case of an open source model.

23:51So the thing that I really deeply believe is that at some point every person will want and really, I believe, deserve an interface that is fully and truly and exclusively aligned to them. But that interface is unlikely to be your whole world, right? The parallel to this is like, you know, I'm speaking to you through macOS. I have a lot of confidence that what is in my Mac, I'm controlling, right? My data is saved to my Mac. I can move that to other devices if I need it. But I still happily use Chrome and Slack and other programs that allow me to tap into that data. And so similarly, I'm afraid of a future in which one is monolithically fully within one experience or application or world.

24:48and really hope that every person has quite a lot of freedom in what they use to help them think or remember or do things. Quickly pivoting a little bit. So, you know, in March, TechCrunch reported on your, on Sam Bart's raise,$23 million Series A, led by Jason and Kindred Ventures. So that's 36 million raised to date, total on top of the 13 million seed from True Ventures up front in Betaworks in last November. Yes, we did a 10 million seed at the beginning of 2025 and then a 3 million pre-seed at the beginning of 2024. Okay, great. You know, there's this old adage, hardware is hard, right? A lot of VCs have kind of avoided investing in hardware companies because it's difficult.

25:32Like there's always a software component, of course, but then you've got the hardware component, which brings logistics into the mix and everything. Has there been anything that's changed in recent years that's made it maybe easier to ship a piece of hardware now than in the past? So there's almost two phases at a very high level. As I see it in developing hardware, there is prototyping and exploring an idea. And then there's productionizing, manufacturing it, bringing it to market. On the prototyping side, things have become much easier. There are quick turn fabricators that can get you 3D printed housings, that can get you custom PCBs.

26:18And increasingly, you know, agents are able to design a lot of that with or for you. So I'm seeing an explosion of ideas and of people trying things out. Then on the productionizing side, certain paths are easier. Some paths are still really hard. So if you're trying to take a product to market, you need to have it really rigorously tested. You need to be able to achieve good volume at good margins. And you need to be able to process all of the logistics, returns, customer support, et cetera. And a lot of how hard that is depends on whether you're trying to make something new and custom versus taking something kind of similar to an off-the-shelf device and then kind of re-skinning it.

27:11The re-skinning approach has become much easier. Is that what you're doing? No. We've chosen the gruelingly painful and expensive path of going for a fully custom design with a really talented and excellent manufacturing partner. But you can do things much faster, much cheaper, thanks to lots of factories popping up around the world that have the ability to work very closely with you and iterate really quickly. the cost of that is you may not be able to achieve the level of customization or quality or feature set or frankly more importantly safety and reliability guarantees that you can by controlling the full stack yourself so we considered that path we looked at a lot of companies that were building smart rings and that could give us something really quickly but we wanted this to be the best possible expression of the device that it could be because we wanted to we wanted In this case, for the full stack, right, throughout the hardware, you know, world, right?

28:15Like there's, you know, Apple, everything Apple does is pretty much full stack, right? Like they design the software, the firmware, the hardware, same with Tesla. So, yeah, there's definitely a case to be made for it in terms of having more control over, you know, future iterations, software updates, things like that. So, yeah, definitely something for startups to think about. I would imagine that they would need to be very intentional about that while they're pitching investors so that they can make sure to get the money that they need to actually see that vision through. Yes, I mean, I think we've been really lucky to work with really experienced and patient investors.

28:54And the green flag I would look for in investors is, do they have a track record of supporting companies in a similar vertical? All of our lead investors have invested in successful consumer software plus hardware companies. Like, I don't know, I ask them for advice like all the time. It's like been really important for where we are. Yeah, amazing. Okay, great. Well, that is a great place to end it on. We're just about out of time. Where can our listeners find you and connect with you online and find out about when the next stream is going to drop? Totally. We're at sandbar.com. We're also sandbar on X and by sandbar on Instagram.

29:39Okay. Well, thanks so much again for joining us on the show. To our listeners, you can find me on X and LinkedIn. You can find Equity on threads and X at EquityPod. Talk to you next time. Equity is hosted by TechCrunch Senior Reporters and produced by Teresa Locansolo with editing by Kel. Subscribe on YouTube or wherever you get your podcasts and find out what's next at techcrunch.com slash events. Thanks so much for listening and we'll talk to you next time. Thank you.