Voice AI with SoundHound

Ann Berry: Hello, everybody. Welcome to Public Live, where we explore the latest trends in business and innovation. I'm your host Ann Barry and thrilled to join you right now. Today's guest is Kayvon Mohajer, Co-founder and CEO of SoundHound, which you guys can find under the ticker S-O-U-N. SoundHound is a voice artificial intelligence platform that helps smart devices, TVs, cars, and restaurants deliver conversational AI experiences to their customers. Kayvon it is such a pleasure to welcome you to Public Live. Thanks for joining.

Keyvan Mohajer: Thank you for having me!

Ann Berry: Okay, well, let's tee this up by talking about the fact that the concept of artificial intelligence, or AI, is something that has been getting a lot of attention, particularly recently with the breakout success of open AI, chat GPT, which is something we've talked about a lot here on Public Life. We can touch on that later in the discussion if you'd like. But before we dig into artificial intelligence in general, would you mind just giving listeners a little bit of background on Soundhound and what the business does?

Keyvan Mohajer: So I started three companies in undergrad and but after I graduated, I decided that I wanted to be a technical founder of a super high-tech company that would make a big impact in the world. So I went to Stanford University, and I got my Ph.D. in electrical engineering. And as I was looking for that, what would that next big change would be? I turned to science fiction, and in particular, Star Trek; I said, what do they have in stock? We don't have obvious answers like spaceships that go faster than light and teleportation and holodecks and replicators. But the one that was less obvious was voice AI. So the ability to talk to computers and robots and devices and rooms and have a natural conversation with them and ask them to do things and ask them to answer questions. And I thought that's something that would happen, for sure. And in my lifetime, and without conclusion, I started working from a dorm room at Stanford with my co-founders.

Ann Berry: Because science fiction does tend to become science fact over time. Right. That's, that's the old adage. So fast forward to today. Kayvon, you're out of your Stanford dorm. Talk to us about how the market for voice AI has evolved and what exactly SoundHound is doing. What does the product actually deliver?

Keyvan Mohajer: We spent the first 10 years in stealth, so we kept to the secret. And we wanted to own the technology and make it better than anything else that was out there, and it took 10 years. But we finally achieved that milestone, and we unveiled it a few years ago. And today, we are one of the few companies that completely owns the full stack of voice AI and conversational AI in-house. So there are a lot of players. But most of those use a platform like ours. And we are one of the very few technology providers. And one of the unique attributes of SoundHound is that we aim really high. And we solve super complex problems that push the boundaries of what is possible. For example, what others are trying to do is they want to make computers catch up to humans in language understanding. But our goal is to make computers better than humans so they can do things that humans cannot do. And we are able to support really complex conversations. And that can make people more productive and make the world a better place. And I'll give you an example. Let's say you are looking for a restaurant and you had Chinese food yesterday. So you don't want Chinese food today. You still don't ask a personal assistant to show me restaurants except for Chinese because you don't think they will work. You kind of have learned the boundaries of where they fail and you just operate in that boundary. You ask them simple questions based on keywords. And if you do ask them to show me restaurants except for Chinese, they all say here are some Chinese restaurants where they actually give you the opposite of what you want. But without technology, you can say here's an example actually works. Today Show me Asian restaurants in San Francisco, excluding Chinese and Japanese, and only show the ones that have more than three stars and are open after 9 pm On Wednesday, and you get exactly what you want. Then you can follow up and refine it, and you have a conversation in a complex conversation. So most of our customers choose us because of our superior tech. But they also choose us because you're an independent provider of voice AI. So we don't hijack their brand, product, user, and data.

Ann Berry: So Keyvan, give us a really specific example of that, to bring this to life for the community. So you just gave the example where voice AI was effectively being used for search, right, that you are articulating, I want to find a restaurant with this cuisine in this area. In that example, who is your client? Is it the search engine is that the restaurant and into what device or into what platform is the user speaking to get that search result?

Keyvan Mohajer: So we power devices and services. So with our cars, TVs, IoT devices, appliances, and those are the endpoints to interact with, you know, these domains. And then we also power services. So, for example, for restaurants. When was the last time you call the restaurant to place an order and you had a good experience, they usually don't pick up the phone. And when they pick up the phone, they are very distracted; it's noise in the background, and they are serving people walking in. And our solution addresses that problem. We take over their phone line; we always pick up the phone so you don't miss restaurants don't miss on any calls. And we give a really good experience for every user. We take their orders, we answer their questions, upsell people and give them recommendations. So we power products and services. And we have this ecosystem vision where we bring them together. So the services that we've voice-enabled now can be accessed by the users of products like you're driving your car, you can order food, you can book appointments, and so on.

Ann Berr: Got it. I mentioned at the beginning that you do service smart devices; you've repeated them: TVs, autos, and restaurants came on to give us a sense of how big your addressable market is, how far you've penetrated it, and how much room there is left to run.

Keyvan Mohajer: The voice AI category in the stock market is estimated to become 160 billion. And if you include search and monetization, you know that search is today keyword based. And Google makes a lot of money for keyword-based because people search with keywords advertisers bid on those keywords. If you include the concept that the next generation of searches conversational AI and comes from advertising, it will be much bigger than that. And there are very few providers because of the high barrier to entry. And today, we powered millions of products. The power of Vizio TVs, for example, they're the second-largest TV maker in the US. Just in the last few years, we have 17 car brands have chosen SoundHound, which is very impressive. And beyond TV and automotive, the power products like Snapchat, and Pandora, we work for companies like Netflix and MasterCard to become a robust IoT (Internet of Things) devices. We announced a strategic partnership with Qualcomm to integrate voice AI into their Snapdragon chipsets and the new Italian service fee analysis that sounds for restaurants that I just talked about.

Ann Berry: Great overview. It's incredible to hear just how far SoundHoud has come from that 10 years and stealth to now having a public company out there and in full visibility. Let's talk a little bit more about AI trends more broadly came on and specifically within Voice AI. Can you talk a little bit more about AI trends more broadly came on and specifically within voice? Can you talk about whether you've seen the market for voice AI tech become a bit more crowded and how SoundHound has kept its edge and a competitive advantage?

Keyvan Mohajer: So I usually, so there are a lot of players in voice because Voice AI is happening, and people are adopting it, and there's a big demand for it. But there's a high barrier to entry. So the core technology providers in voice AI, I will say, a handful at a global scale. I usually say it takes 10 years to create voice AI technologies. And it takes three years to realize it will take 10 years, so I've seen several companies with resources start on the pack and then abandon it. So there's a really high barrier to entry, and as a platinum provider in multiple languages in multiple regions, multiple industries. Soundhound is very unique. So we support about 25 languages. We support multiple industries. We are in Asia; we are in Europe, we are in North America and South America, and we claim to be the best tech. And again a lot of customers are to choose us mainly for the quality of the tech. And the fact that you're independent and we don't hijack their brand and users.

Ann Berry: Could you talk a little bit cable and about how voice AI benefits consumers now and maybe in the future, and just what are some of the more surprising applications that, you know, the average person doesn't really know about yet? What are the hidden nuggets of the use cases that you've seen in your experience?

Keyvan Mohajer: I would answer that question by explaining our three-pillar business strategy because it's very innovative. You know, we've been innovating in the tech, but we also try to be innovative in the business. And the three-pillar business strategy is the answer to that. So let me explain what that is. So in the first pillar, we power devices and products like cars, TVs, and IoT devices. There will be 75 billion IoT devices projected by 2025. So the order of magnitude is larger than smartphones. And most of these devices don't have either the physical room or economical room to add, like a touchscreen or keyboard, or mouse. But the microphone is small and inexpensive. And that's all it takes to add voice to these products. So 75 billion devices, from cars, TVs, IoT devices, and appliances, can be voice-enabled. That's our pillar one. And the business model before that is very proven, you know, we get a royalty. And then we have edge and cloud. In the second pillar, we power services, like food ordering for restaurants and customer service, and think of restaurants as what books spread on Amazon. So Amazon started by selling books. They only did that a lot of people knew Amazon as an online bookstore. And when they did that really well, they did more things. And now they sell everything from A to Z. So for restaurants, there's a very complex problem. We have a lot of focus on that now. And then we'll expand to other areas. So that would be pillar two. So pillar one, and pillar two, are very, very proven business models; they don't depend on each other success of one doesn't depend on the success of the other one. And the third pillar is when we bring them together. So we bring the services that the voice-enabled to the products that be voice-enabled. And that creates a commerce ecosystem. So while you're driving your car, you can talk to your car to control the air conditioning and the radio and navigation, then you can also order food, conveniently and from restaurants that we are both enabled. And this ecosystem benefits everyone. Users, who are the most important part of the equation, are delighted because of the convenience and speed of ordering. Restaurants are happy because they get new leads from the ecosystem. And there's a lead revenue that they share with us. And then, we share that revenue with the product leaders with that car maker with the TV maker. And they benefit as well because, in many cases, they are able to generate incremental recurring revenue for the first time from the product.

Ann Berry: Fascinating; it's been such an interesting part of the debate around AI lately, lots of conversation, talking about how the users of these new technologies are going to have to think perhaps much differently about privacy, or bias, or other ethical topics. When it comes specifically to voice AI and everything that you've seen. What are maybe some of the challenges through that ethical or privacy-focused lens that your space faces?

Keyvan Mohajer: Oh, yeah, so those are big challenges. And they're distinct challenges to privacy is one and bias another one. So we've seen, you know, videos and write the news when you know AI does says the wrong thing or unacceptable things undesired outcome, those happen, and the privacy's and other things like devices are always listening. So on the privacy side, that's one of the advantages that we deliver to our customers: we let them define their privacy policy and let them be as transparent as they want. And as restrictive as they want. In the way, they integrate both AI. And that's one of the disadvantages that, for example, if they get it from the big tech, they just inherit their privacy policy; whatever it is, they can't really control it. So we give flexibility to our customers. And one of the areas that can address privacy concerns is the recognition of the edge. So there's the cloud, and there's the edge so, so you're increasingly seeing both demand from our customers and our ability to run everything on the edge. So, for example, that user's voice doesn't need to leave the device, and then you can have devices that always listen without sharing your audio with third parties. On the bias. That's another challenge. And we are committed to being a pioneer in addressing those challenges. Our approach is that we combine machine learning models and software engineering algorithms to create our solutions. And the software sets the boundaries for the machine learning models. And then machine learning models are companions that make our solutions better within those boundaries that are set by our software engineers. So most of these undesired outcomes, and these are scenarios that you see in the news and other demos, are when machine learning models don't have boundaries, and then they learn from bad data, and then they have unpredictable behavior.

Ann Berry: To wrap everything together, Keyvan, what are some of the priorities that you and your team have in 2023 and beyond were the big growth opportunities that you guys are looking to lean into?

Keyvan Mohajer: So the world has changed. And we are adapting and transforming to the change changes. We are transforming to be a much more financially efficient organization by reducing our costs while still growing our revenue at a very healthy rate. And our goal is to become profitable and cashflow positive within this year. We announced at the end of Q3 of the year, in quiet periods after only us will be publicly disclosed last time end of Q3 we had over $300 million of booking backlog. These are from committed customers. So we have a really solid foundation for our revenue growth with our reduced costs and increased focus. We believe we will increase our revenue growth this year, and we'll achieve our goal of becoming cash flow positive.

Ann Berry: These are big goals, and wishing you all the very best of luck as you pursue those. Just thank you again, Keyvan Mohajer. For everything you shared with us, and your insights, it has been a long journey and really ahead of the explosion in interest. Now that's come SoundHound has been doing it for a while. So thank you, Keyvan from SoundHound the ticker again, folks, is: S-O-U-N for your time today. We will be back soon. Here at Public Live coming to you with even more expert interviews to help you unpack the market and economic trends most relevant to your portfolio. Have a good one, and see you back here next time. Thank you!