Ep. 3: Alexa & The Evolution of Voice AI – with Emerson Sklar

Welcome to Bespoken AI podcast. Press one to start the show.

Hi everyone. Welcome to the Bespoken AI show. This is our third episode and our first interview. Uh, we're incredibly delighted to be joined uh by our good friend and esteemed colleague uh Emerson Sklar, who is the chief evangelist of Alexa. And uh yeah, welcome and uh condolences on being our first interview.

[laughter]

It's a learning experience for us all, I guess.

Yeah, exactly. Uh, but if you could introduce yourself, Emerson, and

Yeah. Uh, I am Emerson Sklar. I'm the chief evangelist of Amazon Alexa, the third in the line of chief evangelists.

Uh, excellent. Yeah, and and that's a that's a great lineage there, too. I'm glad actually you sort of alluded to that. But so I wanted to give a brief history of Emerson and I together. We actually met uh going back, I want to say like 10 years now. I probably should have looked up the exact date, but it was around 10 years. I would guess it was 2017. Amazon had put together partners for the Alexa program, and one of them was Applause. And Emerson was at Applause at the time, which was another testing company uh does crowd-source testing.

And got to meet him through that. Virtually, we started talking about potential business together. Then we met in person at one of the Voice AI summits that Pete Erickson uh organized. Uh, so shout out to Pete there. Uh, I'll send him this clip if I remember. Uh, that was I think in 20 That might have been 2017, too, 2018. But then Emerson joined us at Bespoken as our chief evangelist. Uh, the only time that position has existed at Bespoken. Uh, we had to retire it after he left because uh no one could possibly replace him. But then he went to Amazon and I think he's done spectacular stuff there.

So it's delightful to have him here.

Yeah, it's been it's been an amazing journey.

It has and I mean one thing I just want to mention that is that the person that initially introduced us was Ben Anderson who unfortunately did pass away last year. But I think it's someone that Emerson and I both really dearly love. I know I mean especially you Emerson really close to him but we just I wanted to say that you know we really miss Ben.

Yeah, you know it's it's amazing. Ben besides being a good friend to both of us, a good friend to many in the voice community and really believer in the power of conversational AI. He was actually the person that got me into it. I remember when when Ben and I worked together at Applause we were doing some pre-production testing for Amazon on these weird devices with microphones and speakers that hadn't come out yet. Launched them and he said Emerson I need you to go and buy a whole bunch of these and learn everything that you can about them because I think we'll get to work more with Amazon. He's like but I think I think this is the future. I think you're really going to like it. I was like okay sure I'll you know I'll go and play with them.

And and he was right. He had he superpowers was figuring out what kind of things would be good for good for all the other people around him. That would have been 2014 to 2015 and here we are 12-ish years later and all in on voice. More excited perhaps than I was was even back then.

I would agree. I I think it's caught on uh more than I expected. Uh so when we you and I met we were working on uh the Alexa implementations and and developing that ecosystem uh first with Applause then you joined us at Bespoken and then obviously you've done it at Amazon. Uh we did pivot into the contact center at the spoken, which I did see as sort of an unsexy but necessary business move on our part. Uh, and it allowed us to build a good business. But then, with what's happened now, I never foresaw that this would become, you know, ground zero and such, I mean, I won't say it's the very center of what's happening in AI. I do think that's a little bit elsewhere. I mean, it's probably in coding, but it's generally happening Sure. at Anthropic and Open AI. But I think voice AI is the next most exciting thing happening after coding and those frontier models. Um, and so I guess to turn that into a question there for you, I would say, you know, what what gets you most excited about it in general and then, you know, specifically with Alexa?

Yeah. Um, what what gets me most excited about it in general, I think is the way I I remember seeing an early presentation that Bret Kinsella, another good good friend of ours, founder of Voicebot, I was seeing a presentation that he did where he said that the great things with voice, challenging things, was that everyone knew how to use their voice, natural interface that people have. Um, but if you think about mobile apps, everyone's thumbs are the same. Me tapping on a mobile phone is the same as you. But everybody's voices are different. And from from one standpoint, the the sort of speech recognition part of that aspect is, I think was mostly solved back then and certainly getting close to that now.

But we we all found ourselves teaching ourselves how to talk to computers. Versus what the promise was that the computer was just going to understand us. We had to use sort of stilted language to get it to do what we want. Even in the early days of like Chat GPT, which I credit for really like reinvigorating excitement around this technology, um even then you had to spend a lot of time trying to prompt it to do the exact thing. And what's exciting to me across the industry is that that's that's not the case, and that opens up extreme amount of power and learning for everyone, for people who otherwise would not be able to adapt that way, or would not be comfortable doing it.

My My daughter is 4 and 1/2, her birthday is in a few weeks. Um she's been using Alexa Plus for 2 and 1/2 years. She's using a secret version uh long before we launched. And it's it's totally natural to her. Like I still learn how somebody can interact with this technology when they don't have any of the biases decades of doing it that we have. And that's that is is just amazing. Um and it gets better and better at AI.

Yeah, I would second that. I mean, I've been slower to adopt uh some of the AI tools, but like for example, now I've gotten into and I'm still not using my voice to code, but now I've gotten into, you know, writing code on my phone. You go to a meeting, it's kind of boring, let me send them an instruction, yeah, you know, see if it'll fix that thing up and deploy it out there and you can see see how it's gone. So it Yeah, I it is amazing. Um what is your daughter what does she like most about Alexa now? Like what is she using the most?

Uh well, recently she's discovered fart sounds, uh which is a joy for us to experience dozens of times a day.

Um

Yeah. For her, she loves that it gives her access podcasts and answers to essentially any question that we ever possibly want to know. She's getting better at math, she's getting better eating. Alexa's like a natural conduit for that. I mean, I I remember even when even when was was 1 and 1/2 or 2, she couldn't enunciate clearly enough wake up Alexa to actually trigger the wake word, so we would have to say it for her, but then she would speak and she still was not enunciating particularly clearly, but it was clear enough for Alexa to understand even if we couldn't understand what she said.

And even if we weren't around, she could climb up and tap around on the screen of an Echo Show to get the stuff that she wants. And that's a like that's an amazing relationship with technology that certainly is different than I had when I got my first computer, a Packard Bell, back when I 5 years old or something.

Yeah. Yeah, I mean, that is a huge change. That's It's going to be amazing to see how that generation grows up. I mean, they're they're going to have such an advantage with this stuff. It's going to be uh astounding. It really is.

And that's one of the things to answer the second half of your question about what excites me the most about Alexa. I'm sure we're going to talk about the developer aspect of it, which is incredibly exciting, but for the consumer, um you know, the original vision, I don't know if this is actually true or if this is just sort of like an apocryphal story, but the original vision was that we wanted like a Star Trek computer that is all-knowing and all-powerful and jumps in to help you do something and then disappears in the background. And we're getting closer to that. We're not there yet by any means, but then when when everyone has that kind of power at their fingertips and when it becomes much more ambient and it's everywhere that you want to be, not just in your house, but on the go, the amount of time that that frees you up to the stuff that you really want to do, go bike riding or play with your dogs or swim or whatever, Play video games. Like that's that is as a consumer, that's incredibly exciting for

I agree with that. I think that that that ambient vision generally, I think was a fantastic one. And then specifically, to name check another person that you and I both have worked around for some time, I always and I've mentioned actually this I think I mentioned on every podcast we've done so far, but Adam Cheyer

Yeah.

who was instrumental in the development of Siri, and then was very involved in the voice AI community, and I believe still is. Uh he had such a crystal clear vision of a sort of personal assistant. So, you know, the ambient computing that takes care of everything in your life, you know, uh change my appointment, get me uh you know, get me a meeting with so-and-so, text my wife, all that sort of stuff to just manage your life, you know, get reservations at this restaurant. You know, that was a very determined vision that he had. I think you know, my impression was he felt disappointed that Apple didn't do more to further that.

Um though I don't think the technology was quite there, either. So, they obviously ran into that. But now, I mean I I'm sure you would probably agree with this, it is feasible that you could have that, or at least it's it's so close. And I mean that would be amazing. It'd be incredible.

And what I'm curious your thoughts on this. I I think that it won't Some people perhaps it'll be just one monolithic assistant, their own personal Jarvis that they've built that is just tailored to them. I think for for most people, it's going to be sort of monopoly of uh of different assistants that can specialize in different things to support the human in whatever way is best. And so, like there are things that I would trust Alexa to do, but I wouldn't trust Siri or Gemini or chat GPT, not because those aren't good products, but because my relationship with this one personal assistant and the expectations I have are very different than I would have perhaps with some of the others.

I would totally agree with that. I do think that the the relationship is important. One in terms of, you know, the sort of history that people have with these devices and these companies, but then also just purely the access to data, right? And so that part, you know, I mean, obviously to fill that vision, uh there is that intimate, you know, I don't want to say, you know, access to data. Uh that doesn't make it sound very good or that makes it sound uh that privacy is being invaded, but Amazon does obviously have access to a lot of data there.

I mean, do you see that as an advantage for Alexa Plus?

Absolutely. Um I think this vision has always been the case. Execution and limitations of technology perhaps have prevented it from happening in the past, but Amazon really does see Alexa as that AI connective glue across any of the customer-facing, consumer-facing sort of products, and Alexa already knew a lot of things. Some of these other individual products knew a lot of things about customers. They were somewhat siloed, and act together, couldn't combine to give this kind of 360-degree view of the customer. Again, we're not we're not there yet, but I think we're getting closer. So when you have all of these different signals, and you can use them in a way that is respectful of the user's privacy, that gives them the ability to opt in or opt out based on comfortable with, gives them control of that data, then what we find, and I mean, what we found, still the early days of last Alexa Plus, but people are engaging with it so much more because it provides that additional V.

I remember last year as if funny story I had forgotten that my anniversary was coming up and [laughter] I wasn't even thinking about it. I just said "hungry, lunch" and said "Oh Alexa, what's that sushi place that my wife really likes?" And it was like "Oh, it's it's such and such sushi place here in Seattle. I see your wedding anniversary is coming up. Do you want me to make a dinner reservation?"

Oh, that's a

Son of a that's a great idea. Yes, absolutely. That's a great idea. And it said "Okay, I see on your calendar that the babysitter's coming at 7:00. So, should I make a reservation for two people at 7:30?" Yes. And then it was like "Okay, I booked the reservation on OpenTable. I added to your calendar. I sent you the confirmation email." And it took like 15 seconds or something and that was I mean that was stuff that even for me as a human it would have been hard for me I knew the anniversary was coming. It would have been hard for me to sort of piece all those things together and plan it. It would take me 10 minutes to figure out or something.

Yeah, and I I mean I'm glad you picked on that use case cuz that is one that I have seen demos of probably going back to 2014.

Yeah.

But and and it's not like it was made up then. In 2014 you probably could do that. It just happened that it didn't really actually work for anybody in in the wild and you know, even if you could get it to work for one restaurant or for one person, things hadn't been all pulled together. You know, now that is actually an achievable thing that's happening and I mean pretty amazing realization I would say of a vision that a lot of people have had there.

Yeah, you know, I travel for work all the all the time. Alexa knows what I like to do when I travel. I like to eat drink, like to hike, and I like to walk around cities on foot. And went to Taiwan, which is incredible, I highly recommend. And had a couple extra days, it knew I was by myself, suggested around Taipei for me to walk. Suggested one day that I take an Uber an hour outside of Taipei to a town at the base of a mountain that's entirely inhabited by cats. Then walk up the mountain to this famous shop uh this famous town that is lined with tea shops, which were some of the inspirations for the Studio Ghibli movies I love, which Alexa

Oh, wow.

loved. And then hike down the other side of the mountain through some like really rural um rural housing kind of areas to the South China Sea. And then take an Uber back. That was like exactly what I wanted to do, and it just it was like, "Oh, here's here's what you do." Um but then to like the counter of that, I was in Paris four times this year, but I was in Paris with my wife and daughter uh in around Easter. And it knew they were with me, and it knew when I was working and when I had some free time. And it mapped out stuff that my wife would like to do, and stuff that's appropriate for a four-year-old, like parks and museums and science centers.

It just put the itinerary together, put stuff on our calendar. Like, I didn't have to I didn't have to think about it. I didn't have to do any of that effort. Um just it just did it. Wasn't totally proactive, I have to say. Like, Alexa

Yeah.

what should I do? But yeah, it's incredible. The utility is amazing.

So, I would bucket a lot of that under sort of a personal assistant use case. That's how I think of that, which I think is amazing. Um, what do you think about the use case with say Fire TV? Uh, I saw a recent announcement about that. Uh, I think it's being included for free. I love the whole TV use case with voice. I think it's been so untapped. Um, what do you see as the future there? I mean, and I I guess I want to ask you to comment on a specific dolphin, but do you see that starting cuz like and you and I have had conversations like we've had there's a nice fellow at the NAB, John Clark, but uh, we've talked with who was fantastic.

He sees a lot of potential for it. Uh, that's the National Association of Broadcasters. I mean, I I and I think anybody does. I mean, it's just natural to want to talk to your TV and all the things you could do with it. I mean, how close are we to seeing that really hit scale?

Yeah. Um, the adoption is increasing exponentially. I think it's it is really interesting. I think it's a wise and interesting decision that Fire TV as well as uh, alexa.com on the web that people just get the free access to it regardless of if they're Prime customer or if they're paying for an Alexa Plus subscription. That certainly helps option, but it provides a really interesting way for people to interact with services perhaps that they already used or found and new services that they didn't yet know about. On the Fire TV is a really interesting one because the the sort of standards for a 10-ft device for television have been set for a long time. Same with mobile apps, right? And and websites.

Most mobile apps look and feel about the same. A lot of TV apps look and about the same. Cuz you're interacting with them with an up, down, left, right, and select uh, kind of controller and voice presents another interesting way interact with those existing apps then bringing in the capabilities to interact with Alexa and all the rest of the AI functionality. There's going back to my daughter who I use as my muse for like staying grounded to how you know how non technical voice industry folks interact with this stuff. Um my daughter watches doesn't watch a lot of TV. I don't want to get that impression. But the TV that she watches it's a lot of like PBS Kids shows.

Semi-educational. They have a good story.

Mhm.

Um if I'm in the room I'm generally not paying attention. And so she'll often be like, "Oh yeah, you know, Dad, can we watch that episode of that show where they talked about, you know, rhesus monkeys and they show them in cyberspace."

good.

And I'm like, "I have no idea what you're talking about." And even if she knows the name of the show and we get there to the selection of all the different episodes I've no idea what this but Alexa does and now Fire TV because of Alexa does. So, she can just hold the button on the mic and be like, "I want the episode of WildKratts in this case is the show where they turn into rhesus monkeys and they're wandering around the jungle or something." And it it figures it figures it out. I I think what's interesting though and I I do think it's still the early days.

There There is some There is some like fascinating interplay between something that was originally a voice first interaction or voice only interaction that now can be radically changed on something like a television where you also have real estate but you have a remote as well. I think the Volley folks uh now weekend are doing an an amazing job with that. They've figured out ways to bring things like song quiz and Jeopardy and karaoke and all these things to Fire TV. And where where I'm interested to see where it goes next is then developer tools that we have that have not launched in general availability yet. They're used by tons and tons of our partners, but way that these tools work is that Alexa knows what makes a really good experience on an Echo Dot. What makes a really good experience on a small Echo Show versus a giant Echo Show. What makes a good experience in a car or on a television or something. And customizes action mechanisms, customizes the UI for each of those sort of form factors.

And what I'm and I'm very I'm very curious. I'm not a clairvoyant. Um I don't know where it's going to go, but I'm very curious to see what then that enables when a developer doesn't have to do it. When they can trust the AI and the customer can trust the AI to make them the perfect customized version of this experience on whichever of the dozen devices that they might choose in the moment to interact with it.

Okay. Yeah. Well, that sounds amazing. I mean, if you could have ad hoc experiences created. Um

Not not quite. I mean, you can you can have ad hoc just with an LLM, right? But but more that more that like in in the old Alexa days, right? We I mean, we spent a bunch of time testing to make sure that on a headless device that it worked fine, on a multi-modal device that it worked fine. Uh images would display on a small multi-modal device versus a big big device. Uh one of the reasons we had to do a lot of that was developer had to build that. They had they had to customize it. And not everybody is good at doing all of those lots and lots of different variations.

Yeah.

Whereas now maybe they don't have to do it or they don't have to put in as much time. Certainly the the testing is still I think the test is probably more vital now than it than it used to be. Yeah, it's it's exciting.

Yeah, I mean it's always been our contention and our founding principle that you need more testing as the AI gets more sophisticated, but but it is also getting easier cuz you can just say, you know, try out these different things and it'll report back to you, which is really nice. You know, one of the use cases that John Clark from the NAB had mentioned to me that people were interested in was, you know, brands that wanted to do something or media brands. Um, and I think somebody had actually implemented this. You know, uh your weather person is giving the weather and you can talk back to them and ask questions. I mean, do you think that sounds too much like sci-fi or do you think we'll see stuff like that soon?

Oh, I mean, I think we'll see stuff like that soon. That's not an indication that Amazon is working on any of that. Some someone may have an idea, but um I think we'll see stuff like that soon. I'm curious. I like I I would trust John. John is an expert and a a lovely guy. I would I would trust his insight that people want to interact with it that way. Um, maybe it's just because I'm getting older. I don't I don't know that I personally would want to interact with it that way. Not that I'm the best barometer, but

Well, I liked it. I mean, part of the reason I liked it a lot when he mentioned that. I mean, one uh you know, I mean, he he had some example that somebody put together, I think, for uh weather station in South Carolina and one of the news stations in South Carolina and this is the weather person. And and it it sort of worked. Um, and he was thinking, well, I've got some interest in our parties that say you could do this, but instead have it with Al Roker and not just this person in in South Carolina. And somebody who's a really big name and uh one I like the idea of Al Roker cuz I think everybody loves Al Roker and he'd be fun to talk to. Uh, I'm not sure how long I would make that conversation.

Sure.

But then the other thing is it ties back to a part of my career and a lot of the folks that I worked with in it are now at Amazon. But I was doing interactive audio ads. That's how I got into speech recognition.

Mhm.

And actually Bret Kinsella and I were working together on that on that company. Um, but it was this idea of talking back to audio and or video and the video seems I mean it can Yeah, I mean I think your point is a good one. How practical is it? How much do I really want to do it? I don't know, but there's a gee whiz to it of the idea that you could talk back to your TV. I mean Netflix tried this a little bit where they did the choose your own adventure with

Yeah, with Black Mirror, yeah.

Black Mirror ones, which I I personally I thought it I thought it was one of the worst Black Mirror episodes of that.

Yeah. And and not a particularly good choose your own adventure, either.

Yeah. Yeah, it wasn't one that I wanted to watch again and and explore the different paths on. But yeah, I mean the gee whiz of that is is interesting and I guess also just cuz I mean we're not hyper visual on TV. We all watch too much TV in my household. Uh, so the idea of making the TV more interactive I mean it sounds it sounds like fun.

Yeah, I guess I guess uh, a counterpoint to my own point earlier. AWS I've been traveling all of the world supporting AWS at a bunch of these AWS summits, which has been incredibly fun and super informative because I'm I'm not close to the AWS world generally. So interesting to hang out with their evangelists and SAs and and their customers as well. And there's an AWS team that works with Alexa as if we were customer theirs and go on on their behalf. They have these brilliant brilliant essays on this team. And they built I think we debuted it first at Summit London if I recall.

They had this um large vertical format television that they had an agent running on a Raspberry Pi outputting live 4K video of an avatar. And they had a little microphone kind of stuck into it and you'd walk up and you'd hit a button and you'd talk to it. And you could ask about anything related to AWS and anything related to the exhibit that You know, people would come up. They'd be like, "Oh, how's it going?" You know, that sort of generic question. And then people would say like, "Oh, hey, I sent an email to somebody at AWS and I didn't get a response back, but I have this like deeply technical question.

Yeah. What do you think?" And it would be able to answer them in this, you know, looking like it's a real person. And then it supported I think it was using Amazon speech-to-speech models. So, it supported dozens of languages. And so, you could ask it in any of the languages of any of the places we were going that any of the attendees happen to speak. You could ask any of these languages. And it would speak back to you. Not only would it pronounce the words properly, but the avatar, the mouth would move and the like movements of their face would be as if it was a native speaker of that language. And and it knew that, okay, we're in London.

So, as a British avatar should act a certain way. We're in Australia. We're in Paris. And it it just did all of that immediately. And just watching again the conversations is a very technical audience at the Watching the conversations that people have to do something useful and then sort of have a casual chat because they're enjoying the experience was was uh was fascinating.

And that was just a demo somebody put together?

And that was just a demo that yeah, one of one of these again like uh brilliant nerds on the essay team. It's like, "Oh, yeah, yeah, yeah. I think I can put this together."

Yeah. I mean, that sounds fantastic. So, I mean, there there we go. That's what we need for the Fire TV.

That's what we need. Coming soon, perhaps.

Yeah. Yeah. Um so, you mentioned the developer community.

Yeah.

I guess I'm curious to hear what's next there. I mean, I I saw or

[snorts]

you know, obviously with Amazon and with Alexa, there was a very vital developer community at the outset.

Yep. Over a million registered developers.

And the Alexa champions. Uh I think a very fine history of evangelism. And I'm not just saying that cuz you're on here. I think Amazon's outreach to that community has been exceptional.

Yep.

Um you know, yourself, Paul Cutsinger, David Bitsky, um other folks on the team have always been tremendous to work with. But, it does seem to have pulled back somewhat with Alexa Plus. And I've seen that you know, there's bigger names now that are being onboarded to Alexa Plus. But, what do you see as the future for I don't want to say this Well, I mean, yeah, smaller guys or you know, for everyone else. I mean, enterprises that are not necessarily because it's also the ones that have been onboarded are things like an Uber, a Grubhub, things that are strategically um you know, see obviously strategically valuable to the platform.

You know, there's lots of enterprises that may not be quite so strategic. What's the future for them, do you think?

Yeah, a bunch of questions to unpack. Um So, I I think first, the customers have an expectation of but uh hyper-technical people and stuff aside, but like, you know, regular lay people customers have an expectation that the system is going to be able to do certain things. They have an expectation that Alexa is going to be able to do the old stuff that the original Alexa, as we call it, has been able to do for a decade that it's going to continue to be able to do it. The expectation is going to be able to do new flashy things, some of which may provide some value.

And they have an expectation that it's going to do thing easier, faster, maybe cheaper, scalable, or actively that it's going to be able to do something slightly differentiated perhaps than what they could do if they opened up a website or app on their own mobile phone. And like I think think that's one aspect of some of those large enterprise partners that see that we've launched in the US and the other I think we're in 11 countries right now. Um I think that's I think that's one aspect of it. I think a second is that the state-of-the-art not only continues to evolve, but it evolves faster and faster.

In July, I believe we announced new SDKs. And new SDKs at that point were in the hands of our enterprise partners. We're using them internally. We're actually using all of those same all that same tooling to build uh internal Amazon functionality that Alexa does, the weather and timers and stuff. And that stuff was so drastically different and so cutting edge compared to legacy developer tools, say. And they enabled people to go at people who are not voice experts, but are experts in some domain or some specific type of application or vertical or whatever, enabled them to build something that worked on Alexa better than any comparable thing of that type has ever worked before.

Flash forward to about a month ago, we announced some different SDK stuff or an evolution, let's say, of SDKs. Um still have much of the same capabilities, but we've fully embraced MCP as an easy way for people to connect whatever it is that they have, whatever it is that they do, to Alexa. And just like before, the ability to make something that works better than you could possibly fine-tune something to work on the original Alexa and to do it in no time at all is is mind-boggling. I'll give you an example. I remember watching I don't remember if it was Dave Isky or Jeff Blankburg. I remember watching one of them in the early days of the Alexa skills kit. And they they all used to do like live coding sessions. Yep. Yep.

Yep. And I remember watching one of them spending weeks working on a Pokédex skill. You can query about any of the information about Pokémon. And I was using uh today I was using this new MCP toolkit that we announced about a month ago to build a Pokédex. You know, to see like what it would be like. And I'm using a genetic coding tools. It would have taken me longer certainly if I was trying to code by hand, but how much longer? And I built a rich Pokédex experience um

[laughter]

in like 20 minutes in the background of a boring call that I was on. Um and it works beautifully and I can say like, "Oh, what happens if I have this Pokémon fighting this Pokémon?" And it like intelligently reasons across all of it. And I I only tested it on like a simulator and on a device, but I could have gone to alexa.com and tried to interact with it there. I could have pulled out my mobile phone and interact with it there. I don't have a BMW anymore, but if I had a BMW that had Alexa in it, I could have hopped in my car and tried to test there and that would have taken me months to do otherwise and it wouldn't have worked an ounce as well as it just this morning.

And that that pace of innovation is is amazing. It's amazing to be part of it. Amazing to be part of it as a technical person. I think we have a great responsibility to the millions and millions of people out in the world that the almost 700 million devices that have Alexa on them that they use every single day. I think we have a responsibility to them to make sure that we are building, make sure that we are deploying to them better experiences than they had before.

They're trustworthy, that work really well, that are reliable, that are engaging. And we have a responsibility to our developers and partners to do that, too. And that is for better or something that just even in the age of that the coding something that just just takes time. Um and you can ensure with some a select group of giant enterprise partners that have a you know big engineering wing. Um but we got to make sure we only have one chance to get it right. We've got to make sure we get it right. Um but the the like long thread behind it, I am I am out I guess I don't know honored and pleased to see or pleased to hear that time and time again every conversation that I have with Panos who is the head of all of Amazon devices, he's a member of the S-team, he reports to Andy Jassy, that Daniel Rausch who is the head of Amazon Alexa who reports to Panos, that the leaders underneath him that their peers, that everyone has everyone continues to have a unshakably strong conviction that enabling a broad scale all three P developer like not enabling the broad scale developer community to do this unanticipated evasion to bring the stuff that they do best to the clever ideas that they have into Alexa and then out to the millions of people that that is absolutely vital to Alexa.

And I think we just want to make sure that we make it as easy as possible and make it so that they can ultimately be successful in the way that they invest their time and engineering effort.

I mean I think that's really well expressed. Um yeah, I mean I want I just want to say that I I appreciate the the sentiments you put out there and it's it's nice to hear that uh that concern for the developer community. I guess one question I would have about it though is I mean do you think the developer community I mean is there as vital a role to play because these things can do so much now? You know, I use Gemini for most of my sort of just like typing stuff in asking questions, right? I like Gemini a lot. It sort of most feels like it's I don't know. It gets the job done for me. But I mean I don't feel a need to go outside of it, you know?

It does have such a complete uh access to all the sort of information I'm looking for um all the capabilities I guess that I'm looking for um I mean I'm not using it to control my email, but I you know, for so many queries it is so self-sufficient. I think this is true of all the frontier models now. And if they're not doing something, it's you know, it's easy to imagine that it would be soon cuz we've obviously seen that. So, I mean, does that weaken the position of the developer community or for the, you know, the vision of a very large group of people contributing?

It's a fair question. Um if you know, there's there's a lot of just like inherent capabilities of large language models, let alone uh sophisticated 80-plus model orchestration assistant like Alexa Plus that that if it were an apples-to-apples thing, if I say uh I say tell me a dad joke or something. All right, like every LLM in the world is going to know every dad joke ever written by mankind and it's going to be able just to recite them like them ad nauseam and um would could something like that in an apples-to-apples world, could something like that succeed, let's say?

Um my gut reaction would be no, except except that there are, looking at the data from Alexa skills, for example, even with people that are on Alexa Plus, there are such a huge volume of different types of experiences that as someone who is not actually a user of them, I might say like, oh, I think I think an LLM could probably do that, but people refuse to use an LLM to do. And that I think is an indication of both the tight relationship that a company, an independent developer, whatever, tight relationship they built with with a large body of consumers.

But then a specialization that as the experts in whatever that thing is that they know that Alexa doesn't. So I think I think A, there is a huge opportunity for perhaps it's got to be some sort of differentiation, but for people to have that differentiation even if it's a calculator and I say what's 2 + 2 and the Alexa one's going to give me four and the custom one's going to give me four. Like I think there is still there's still opportunity for differentiation there. But then it's it's this notion that they had when the Alexa skills kit launched of this unanticipated innovation of stuff that the thousands of people that work on Alexa don't know and not clever enough to think about I guess.

Um that makes that presents this incredible opportunity for for folks. I mean again I'll go go back to uh fart sounds. Um It it is entertaining to me to hear my daughter giggle every day when she plays fart sounds. But if you had asked me in 2015 if so many people would be so equally entertained by fart sounds and that it would be such a durable experience that people still love. You know I I wouldn't have foreseen that.

[laughter]

That is unanticipated to me. Um

Yeah.

And and again when I can go today and like the thing that I made today I mean it was like you know it was a proof of concept. It was like the modern day version of a hello world kind of thing. Um but it's amazing and when somebody like me can do that so rapidly it means that you could do it in the exact same amount of time. Other people could do it. We would all end up with the same sort of middling kind of experience that is probably more representative of the latent capabilities of the LLM then our own creativity. And that then I think presents an incredible opportunity for somebody to come in and make something mind-blowing that then next generation of Alexa users, which now more different people are using Alexa than ever used Alexa before, and even the people who use Alexa a lot in the past are using it more every day.

Like that's that's an an incredible opportunity. I don't have the answer. Like I'm I'm admittedly not that creative in that regard, so I don't know what it will be, but even even with things that are LLM latent kind of capabilities, they'll prefer is interesting differentiated experiences. So, I think it's a it's still very much an open playing field.

I love that note. I think that's I think we should actually wrap up on the note cuz I I like how optimistic that is, and it also just takes me back to when I was starting Bespoken, and even when we started our professional connection, you know, it was based and and I think it was less realistic then. We did believe there was going to be millions of apps, but I really wasn't sure how. And the limiting factor for me was just like I wasn't sure about some of the technical limitations. And I think those technical limitations did prove in that round to be you know, I mean they they did stymie some of the progress. And those are gone now. So, I think that you know, the only limiting factor is people's creativity at this point, and uh yeah, I wouldn't bet against that. I think it's got immense potential.

Yeah. Yeah, I think I mean I I was excited back then. Um I've been excited 12 years or so, but that is one thing that keeps me up at night is like how do we how do we achieve this incredible like potential and dream and then it gets me out of bed in the morning and gets me glued to my laptop again um seeing the progress and seeing just the the possibilities and being curious of what it's going to look like a week, a month, a year from now.

Well, I think that's a good note to end it on. Emerson, thank you so much for joining us.

My pleasure. Thank you for having me.

Yeah. Yeah. It's terrific to talk to you and uh you know, we look forward to seeing what comes next with Alexa Plus and everything going on in Amazon.

Sounds great. Yeah, I look forward to more updates perhaps the next time that I come come here on the pod.

Yeah. Sounds great. Thanks.