Ep. 1: Voice AI News & CCW Wrap-Up – The Bespoken AI Show

Welcome to Bespoken AI podcast. Press one for the latest news in voice AI. Press two to start the show.

Hello, welcome to the Bespoken AI show. We're building this podcast to dedicate to the news and history of Voice AI. My name is Diego Sandoval. I'm the head of business development at Bespoken.

And I'm John Kelvy. I'm CEO and co-founder of Bespoken.

And in this few episodes we're launching, we're going to be covering a variety of topics. the developments in voice, AI agents, and we have several headlines to discuss, but we're also going to be covering what we're learning along the way, some takeaways that we have with customers, our engagements with conferences. Uh, but to kick things off, we want to talk about what's been on the news this past week. And the first one on our list is Salesforce acquiring Finn. For those of you not familiar with Finn, it was formerly known as Intercom, and it was recently acquired for $3.6 6 billion to enhance Salesforce's agent force platform. For those of you not familiar with the intercom uh core product, you're probably seeing these widgets on the websites that open up a chat either connecting you with real people in their team or with a custom agent that is resolving your issues. Uh so lots to talk about on this. Uh I'm curious what your thoughts are in it, John.

Yeah, I mean I think this is a really interesting story. Uh this is at least partially sort of an inspiration for the show because when this happened I was practically kind of grabbing people on the street to discuss it. I thought this was a really interesting acquisition on the part of Salesforce. They're not traditionally someone who is you know closely associated with voice AI but they're definitely a company that you know I watch and I think most people watch uh and certainly has made a lot of noise about AI you know and and has been touting their agent force for some time.

Uh so then when they acquired Finn which we knew as intercom and used as intercom I thought that was pretty interesting and and really there's a few parts to it that I thought were interesting. One is obviously just the part of where you know it does look a bit redundant. So you know Asian force we've heard a lot about it. I think certainly part of it is it's meant to be used for customer support and so is Finn. So it's like well why did they need this? And I do think part of the story there, you know, or that part of the story, I think a lot of people had questions about that. I looked at the stock price for Salesforce, uh, it's been, you know, a little bit less than 30 days since this acquisition was announced and it's down 20%. And I don't think that's just the Finn acquisition, but that's a pretty remarkable loss of value given how much Salesforce is worth. Uh so I do think it did shake people's confidence that Salesforce knows what it's doing from an AI perspective. I guess another aspect of this though that I would say that's perhaps a bit more optimistic about it in terms of what Salesforce can get out of it is you know what you hit on there Diego which is you know Finn does have this sort of valuable real estate in terms of interacting with end users and so you know agent force maybe ABA can do all the things that Finn does but it doesn't have that immediate reach into end users uh which even though it's relatively easy to install these products, the fact is getting onto everybody's web pages, I do see it a bit less often now, but getting it onto everybody's web pages, you know, that's a non-trivial thing. I mean, that that does take time and so it then becomes this beach head for Salesforce to presumably push agent force a lot more and to push their AI strategy a lot more. Uh so I guess, you know, that part seems a little bit more optimistic for Salesforce and and to me helps make some sense of it. I think as a goto market uh adoption I think that's a good point. Uh I think it was something around uh I don't want to butcher the numbers but several hundred,000 customers that Finn is bringing into Salesforce and where now Salesforce can deploy their capacity into those front-end pages. And I I would like to entertain uh just for a minute before we move on to our next headline uh the story of whether this is an act of desperation uh and what context is in there. I think it's worth noting that there was an opportunity to acquire uh Sierra uh which was started by a Salesforce executive Brett Taylor.

They've been growing massively in the last three years, acquiring a good chunk. I think it was about 40% of Fortune 500 companies, if not Fortune 50. I'd have to look into that stat again, but just rapid growth. And for whatever reason, they didn't acquire them. So, was this just the next best move they could make?

Yeah. I mean, I'm glad you brought that up because I think well, one, yes, I think that this does look like a little bit of a consolation price because they were not able to acquire Sierra or reach some sort of agreement to merge or tie up. It does look like Sierra has been so successful at fundraising and getting early traction with customers that, you know, that no longer really made sense for them. And you know, I don't I don't know if Marc Benioff thought that that was sort of always gonna be an option for him or it was going to be an option for a bit longer and then it sort of got away from him. I know, but it's, you know, Brett Taylor has a legendary reputation at Facebook, at Salesforce.

He's on the board of Open AI, you know, I mean, he's he's a luminary, I would say, in the technology field. I think almost everybody would agree with that. And so that looks like a bit of a missed opportunity for Salesforce. I would also say that, you know, I I think it's interesting to bring up Sierra because that is a subject, you know, there's certain themes that we want to return to as we're talking on this podcast. And I think Sierra is definitely one of the companies that we have our eye on. We don't have a news story or anything to talk about them this week, but you know, to me, they are in this new firmament of Voice AI. They are a major major fixture. What they have done still at a very early stage. I don't think you know they're gonna by any means just run away with enterprise AI. I mean it could happen but it's it's hardly written in stone at this point.

But they've taken a great early lead. I think they've they seem to have done a tremendous job. And I guess it's unsurprising given the reputation that Brett Taylor has. I mean he seems like a very capable executive. It's going to be really exciting to see, you know, the next couple years where Sierra goes. But I think that's a good segue for our next headline. Siri AI. I know you're a fan of Siri. I haven't really honestly adopted Siri. I only trigger it accidentally on the iPhone, but I'm curious to know I'm curious to know what's new. So, what have you seen? So, I was really excited about this announcement and even from the time that we, you know, you and I did some initial uh sort of runthroughs getting ready for this. I've read more about the announcement to try to understand really what it was. And I think it's a little vague and that maybe because the last time they made a big announcement about AI and Siri, they got a big class action lawsuit and uh you know, I mean, lost big on that. That was fairly remarkable.

So this one I think you know they're sort of indicating there's big things coming but they're being a far less specific about what those things are. It's more like sort of general capabilities. It'll be able to see what you're doing and enhance it. It's going to be more front and center and then it's going to be in that dynamic island. There's going to be an app. It's going to have built-in I think the one specific thing I saw was you know it's going to be able to help you with your writing wherever which is like okay. I mean, that's I mean, that feels a little 2022 at this point, but still, I'll you know, good. I I'll take that. I mean, I I need help with my writing. But, you know, overall, I I think there's reason to be optimistic that this will really bring a lot to the table and that we will see a lot of innovation here. I think certainly Apple is paying more attention to it. And I sort of expect this to be not triumphant, but I have heard for the last few years what is wrong with Apple's AI strategy and Apple's not doing anything and they're not being effective. And look, I mean, definitely the last announcement with Siri was a misstep. There's no question about that. I mean, I don't think that looks strategic at all, but Apple has a long history of letting people fight it out and then they come in and they just take the market. And I think that could easily happen here. Uh they have perhaps the best channel to end users certainly one of the top ones. You know them, Google, Facebook have, you know, sort of these most obvious touch points with consumers. But I think Apple particularly is in a great position with the iPhone, you know, the earpods that I'm using. Uh as well as their emphasis on privacy. that they have this ability to kind of pull a lot of things together with AI in a way that other people don't and to guarantee usage. You know, assuming people are not completely annoyed with Siri. I mean, you sound a little annoyed with it, Diego, but you know, assuming people have some goodwill there, I think it has a chance to really be successful and and really drive a lot of usage for them.

I do want to spend like a minute or two more on on Siri because to my credit, I haven't really fully leveraged it. My use case is to yell out, "Hey Siri," when I lose my phone and it it kind [clears throat] of responds to me and I I find like but going back to like the vision of Siri is really is a personal assistant and so I have been using you know chattude as personal assistance to go run tasks for me. I have not given that responsibility to Siri. I'm curious if that's the direction it wants to take now that these new features seem to allude to the ability to navigate different apps, go to different parts of your of your phone, maybe pull up like an old photo, create a calendar invite, do things for you, right? So, to that note, kind of two notes there. One, it's got the real estate, and we've talked about how Finn positioned itself on the customer website. So, we're touching on a on a theme here of like where does the agent live and what leverage is that giving companies and has Apple fully leveraged or taken advantage of the real estate they have or are they waiting and then the second point the use cases and I'm curious how are you using Siri and how do you expect this virtual assistant uh or personal assistant to evolve?

Yeah, I mean I I like what you said there about it being a personal assistant. It obviously has not lived up to that so far. I will say I actually love Siri even in its current incarnation. I know it's limited. I spend so much time in this space and I have spent so much time in the space that I have heard all the criticisms and certainly I I acknowledge like it it seemingly could do a lot more especially for the past few years. It does seem like a little bit of a missed opportunity. But even in its current form, I mean I do use it to play music.

I use it when I go for bike rides. I like to take notes on audiobooks that I listen to, which I think is just a fantastic use case. I mean, I sort of bike around talking to Siri, saying [laughter] things about what I just was listening to. And yeah, I mean, it's I I think it's fantastic for that. And it's easy to imagine it doing so much more, you know, and it does take me back. I I think this is something that maybe we'll go into more depth on in a future episode, Diego, but the sort of history of Syria, I think, is fascinating. And I think particularly Adam Cher, who led that team for a long time, you know, it had its roots at the Stanford Research Institute, which is where the Siri SRRI comes from, went over to Apple.

Adam, maybe we can put a link to this in in our notes, but Adam, I mean, you can go back many years. He has always had this incredibly crystal clear and I think amazing vision for what Siri should be and what a personal assistant should be. And the idea of something that, you know, you say, "Oh, I'm running late, so text my wife. I'm I have an appointment conflict, so let Diego know and let's let's move our onetoone to a half an hour later. Make an appointment with the dentist. Get me tickets for the theater, but don't spend more than 70 bucks." Like I mean he's had these use cases I feel like completely galvanized in his mind. You know some of it is kind of obvious because it's it's essentially we'd all like to have a personal assistant and unless you're like Elon Musk I mean you don't you know but there's a lot of stuff that seemingly these things could take care of for us that would really make our lives easier and you know there's a lot that AI does but just the personal assistant aspect I think Apple is uniquely well suited to deliver on that. They have all the data. They have the trust from people by and large. And you know, the technology certainly seems like it's there. Not quite sure what we can expect in this next release, but it would be easy to imagine maybe not everything I was just describing, but that you're getting closer to that. And I I think it's going to be fantastic.

And it would also, you know, does touch upon a second theme that I think we'll return to again, which is what are the channels? How do you get access to people with AI? And as much money as Claude and OpenAI and other companies may raise, as much of a great position as Google is in, owning that phone experience, to me, it's a uniquely valuable piece of real estate that, you know, even if Apple has spent nothing to date, they can easily run away with this and be a huge, huge winner. Uh maybe the biggest winner in consumer AI just thanks to having that existing relationship.

I I think that was really insightful. I can't mention this headline yet because we I haven't looked into it, but I saw Elon was kind of flirting with the idea of launching his own phone, but that that would be a really good next topic to tie that back in. So,

yeah, I just what he'll call it alone. I mean, there's probably like four or five stories there that are worth talking about. I mean, the naming

Yeah. So, fascinating.

We've got another one on our list, and this is sort of going back into the contact center realm. So, Amazon announced a new UI for their connect. I think one of the key takeaways here is that it's catering more towards business users as opposed to your typical engineers. So, there's an emphasis on low code, no code, uh helping non-technical people uh get to production faster. So, I'm curious what your thoughts are on that. So this I do think gets more into our core competency as a business around you know helping people to deliver great voice experiences because yes I think this this was part of an acquisition that Amazon did. I think it was you know I think it was a good acquisition. I think this is going to be good for Amazon Connect users and I guess you know the sort of things I've seen from Amazon Connect in the last six months. There's been some other news stories and they seem to really be pushing it in a few different directions rather aggressively that I think are, you know, make sense to me. I think that's all good. At the same time, I did go back and I look through the history of no code tools being announced as part of Amazon Connect. And I mean even with a brief look back you can find in 2022 they were touting Lex being no code and Amazon Lex had you know I think it even going back to like 2018 or 2019 that was being integrated into Amazon connect as an simpler way to build things maybe more orient towards developers but still had a no code sort of low code story there but in 2022 they're sort of really emphasizing that part because they added more uh well they basically added the workflow diagram type tooling that you know many people are familiar with. Uh then they went in 2023 introduced that functionality into connect which at the time I was kind of scratching my head over that one because it looked you know it's like why do you need two workflow tools and connect was by no means a sophisticated NLU. Um, arguably Lexus was not either, but you know, there was some NLU capability. It seemed very duplicative. And then with this announcement in 2026, it does sort of seem like, look, I mean, you guys have frittered away some time here. And I do think, you know, when the AI push was really starting, I mean, when Chat GBT was obviously such a major event when that came out and sort of changed the world seemingly, I really thought, well, Amazon's going to crush it with Connect.

And for the same reason I was just saying with Apple because it has such a great channel. You know, Amazon has or in my perception it had great AI and then they have this great channel in a CCAS contact center as a service with Amazon Connect. And so you marry those two things together and you have this incredible ability to deliver new AI experiences to end users. And I do think having that channel and the technology together is going to be, you know, when I think the sort of growing pains that I anticipate for our industry in the next few years, just that simple equation, I think is going to be such a barrier. But I didn't think I thought, well, this is where Amazon's going to crush everybody.

It seems like they're kind of spinning their wheels. And it's uh it's kind of surprising. you know, Google, they didn't opt to build or buy a CCAS. They didn't seem to want to do that. Uh there's a lot of CCaaS providers that we know well, you know, the Genesis, 59, Nice, other players that are out there. They don't have the same history with building out AI solutions. And I'm not saying that they can't. They're all working hard on it. Nice made an acquisition with Cognagi that was impressive, etc., etc. But nice did not have as Amazon did 10,000 engineers working on Amazon Alexa, you know, working on a voice AI product in house that seemingly could be very quickly repurposed to deliver generative AI uh into these experiences. So I mean the channel story to hear to me is interesting but I think also you know there is an aspect and I want to phrase this I guess diplomatically because I don't want to sound too critical but there's a part of it where I feel like Amazon squandered some advantages and even part of me where I think Amazon has been the best technology company for 16 odd years running. They've delivered technology at scale via AWS far better than anyone else out there. you know, has their model sort of met its match with AI? And without going on too long here, I'll just the last thought is I think what has made Amazon so great as a technology company is they're super customer focused. Everybody knows that.

And they hire incredibly bright people and it's just like let's solve these problems. But, you know, as opposed to somebody like Google that is definitely not as much customer focused. Uh I think some people could say it's almost customer allergic. uh they don't really like talking to people sort of famously. I don't think I'm being overly critical to say that, but they are much better at R&D, right? And so is the lack of emphasis, the lack of culture around R&D at Amazon sort of holding them back? That's a theory I have, but either way, it does seem like they've squandered a little bit of an opportunity here in 2026 to just be bringing in a technology uh NLX, which was a very wellrespected company, but you know, it's not Frontier models. is not a frontier technology. It it does look like they've missed out a little bit.

That's really interesting from my perspective. What I'm seeing as well is this when you cater to a new end user, let's say the business analyst and giving him the tools to run with the project and not rely on engineers. I've seen the story on a bit of a different angle more from like an app development, you know, people building apps. We've seen the vibe coding wave. uh first we saw the promise of visual development. So you had platforms like Bubble which let you put the logic pieces and then build a a UI and a back end and you've launched an app and then that got steamrolled by Vibe coding uh like lovable uh you just prompt

your app idea and we've seen the promise and we've also seen the reality of you know you've given this superpower to people that are not trained to be engineers and they suddenly discover security issues and I've seen that story play out in that world. I'm curious how that plays out in the enterprise and what are the sort of wake up or reality checks that happen in the next few years when we do democratize access to building these systems. I mean, I'm I'm curious whether you've seen sort of that trend as well.

I think that that's a great question. I think that that's one that hasn't been answered. Well, well, so we do have some early returns on in the sense on the chatbot side. My impression is that they were just sort of turned loose. I mean, there was a wave of innovation that sort of transformed everybody's chat bots, everybody's customer support bots, and it happened. I mean, I'm not saying everybody has replaced the bots on their website, but the quality of answer you now get from those little chat bots that live in the same place as Intercom did and a lot of times still is intercom or it's a new type of widget or it's, you know, I mean, you're going to a separate website for customer support. Whatever it is, the chatbot genai rag model experience is amazing. I mean, things that would have been tickets or hours of searching through help documentation are now resolved in about 10 seconds. And so that I think was transformative. And at the same time, from where we were sitting, we didn't see a lot of concern around does this work right? Does it ever make mistakes? you know, most people seem to be very tolerant of errors with it as long as the containment rate was up, meaning that, you know, end users were able to resolve their queries uh without having to talk to a human being. And you know, presumably, I don't know how closely people were watching it, but presumably that the NPS scores weren't sinking, which I don't think they were. I mean, like for many of the people we work with, you know, my delight with these products has gone way up because you don't have to open a ticket. You don't have to wait for someone. You don't have to look through tons of documentation.

And I do see it sometimes, not not that often actually make terrible errors, but yeah, it doesn't seem to bother anyone. That's one result. Will it be the same for other applications of AI? Because, you know, these are basically glorified FAQs and documentation bots. I don't know.

That's interesting. I I think the pace of, you know, they're going beyond rag now with the MCPS, they're going out and doing things. It just opens up a whole layer of things to to consider. So that's a great segue actually to this next headline, ElevenLabs and the resurgence of procedures.

Yeah, I love this story. I mean, ElevenLabs announcing their new product procedures, which to me, I did a lot of work on this. I actually had a company around this. It looks like a rules-based system. So it is a structured approach to encoding policy in plain language. And you know, I think it's a good way to encode policies. It's a good way to encode guard rails. It seems like uh I don't think they've been adopted necessarily at scale yet, but it seems like this may be, you know, to the question you were just posing, Diego, for things that are more sensitive, for things that have more of a sort of regulatory overhang, you know, I mean, if you deploy a chatbot that answers questions about how do you use an API and it doesn't answer right, big deal if it gets wrong. If you're a bank deploying a customer support bot and it's answering a question about somebody's mortgage policy or uh if it's making an underwriting decision about whether or not somebody should get a loan, if it's an insurance company deciding whether or not somebody should be insured or whether or not something is covered. These are questions that presumably well not presumably but they have far more implications legally as well as reputationally for companies and rules-based systems when I worked on them I don't want to say when but it was in a previous previous wave of AI that was less successful you know they were seen as a natural solution because the problem people were always trying to solve was like you know we have these complex policies and we want the people that are the experts in the policy to be able to manage it and this procedures looks like exactly that type of tool. Let's let the sort of subject matter experts own the policy and I think to give Amazon kudos to for something because we're sort of being a little bit negative towards them in that last work I will say they put out a great paper on an approach to turning policy uh essentially turning it into mathematical logical expressions. Uh we should provide a link to this too. I'm not going to do it justice in terms of summarizing exactly how they're doing it, but you know, essentially people the gist of it is express your policy in plain English, leverage generative AI to transform it into a mathematical equation, do testing on it to ensure that that transformation step was correct as well as that, you know, I mean, having worked in rulesbased systems, one of the things that we learned and like my partner who's former CEO of a major mortgage company can can talk about what you'd find with these rulesbased systems is that the policies were in contradiction to each other. They could you know for 20 years people could have policies around what what your underwriting is, what your securization is. You know banks could live with these policies that were in direct contradiction for each other. But you wouldn't know that because you know it might be thousands of pages you have to go through to find those contradictions. And so it wasn't easy to identify them. Uh, so that's something people have to kind of work through as they get to that. But the direction to me makes a lot of sense that you have an LLM that can answer things but then is supervised by a rules-based system. And that's what this procedures kind of looks like to me. And I think it's cool. So I'm really interested to see where this goes. And I'm also just grateful that in the sense that it it sort of makes some sense out of part of my career that to me looked like kind of a weird detour. It definitely some people would tout rules-based systems as AI at the time and I was like that doesn't seem that smart. So hopefully they'll be smarter this time around.

I could keep talking about that. Uh at some point maybe in a future episode we could dive deeper into what lessons or obstacles could you see in that wave of AI happening or resurging in this wave and are we in a bubble and go into that but but it'd be it'd be interesting [laughter] to save that for another time. I feel like the bubble should be on a list of verboten subjects.

Yeah.

Not because I don't like it, but I will say that I if you have a beer with me, I like to make predictions about whether or not we're in a bubble. But I mean, I've just heard the bubble conversation so many times. It's like, yeah, I guess it's hard to say. Regarding ElevenLabs, I will say they are another company and another theme that I think we'll be returning to. They along with Sierra to me have done this great job of sort of getting out there. I haven't seen a lot of customers yet using their stuff. I know they do have great customers, but I do think it's it's somewhat early for them, but just in terms of how they've shaped the conversation, their influence in the market, the sort of mind share that they have amongst developers and business users, I think it's amazing. I've enjoyed listening to the CEO of ElevenLabs when he's appeared on podcast. Seems like a very sharp guy. So, I I thought this was a great announcement. It's it's a little bit technical, but I think it's a good one. It seems like a good product release. And um you know, we're interested to see where it goes. And I guess one other thing that will kind of tie this into a final point that we're making, but I think this also this enables testing. So, one thing I've been worried about from the very onset of LMS and generative AI is how do you test it when the behavior is so non-deterministic? Well, if you have a policy encoded in natural language, you know, that is explicit that does express these are the constraints on the system, these are the things that it should do and it should not do. Well, that becomes your test bed and that becomes a tool that we as a company that is built around QA, you know, we can really go to town with.

So, from that perspective, I think it's also a great boon. uh it makes these things that have seemed a little scary, a little non-deterministic and potentially erratic. Um it holds the promise that they will be much less so. So I I'm really interested to see where this part goes. To me, this this points the direction to the future for how this stuff really moves into production uh in the sort of highly regulated industries that we work with. And I I think it's also a good uh segue into the ElevenLabs announcement made at CCW that I saw last week. Uh they had a partnership with Telus Digital and the use case was very interesting. If I understood it correctly, it was about using ElevenLabs to simulate calls with agents that are being onboarded.

Oh yeah.

And ramp up that time that it takes for them to get the training they need to go out and face real customers. That's what I understood as as the use case there. And [clears throat] so in that sense, they simulated, you know, human bot interactions, conversations, and basically sped up that that onboarding period by 20%. So I thought that was really interesting. Uh it's also just one of several use cases that we've heard about coming out of ElevenLabs. So very interesting to see that.

Yeah, I heard about that one from some folks at Telus. I think that's really cool. And that's, you know,

that's one of those use cases in AI where you're just like, I never would have thought of that, but it sounds great and, you know, sounds potentially like it's a lot better than what people might have done before.

And on CCW, there was just a lot to talk about. I mean, there were about 300 booths, so exhibitors all involved whether in CCaaS or in workforce management. a lot of agent assist as well. But on the voice AI space, uh what I'm hearing uh what I've seen is just an emphasis on, you know, getting to production, no longer staying in the pilot phase. You know, that that's been sort of the drum being beat this year. And at least on the QA perspective, we have a firsthand look at that. Uh and still we we do have cases where we've seen projects being invested in and then having to be cut because it didn't meet that quality threshold that that gives them that confidence to launch. So

I can't say that I've I've been able to pinpoint any success story as of yet that that's compelling but it you know there just seems to be a lot of enthusiasm around getting to production and people are really taking the guardrail conversation uh seriously. So I would say a lot of the programming was around guardrails, governance and I don't see QA being explicitly said but you know in the in the context of these procedures and and making sure your agent is following some set of rules you know that that seems to be the major trend.

Yeah. Yeah. I mean that's interesting uh that you say that guard rails were a big point of conversation. I mean who seemed to be sort of moving the fastest? I mean what was the sort of most innovative vendor or use case that you saw on display.

I personally thought Sierra was very compelling, very full stack. It was one of the largest booths and I guess their sales team was really well-versed, very enthusiastic as well. So they, you know, you left their booth feeling like, wow, they are here for the taking. And then

yeah,

you know, much different energy than when I would go to, let's say, a Genesis. Genesis was the the other one that was like took up the majority of the expo hall and they were doing lots of demos as well. But what what I would highlight is you know of all these I would call them like contact center AI Sierra and ElevenLabs too but there was also another one the name is escaping me right now but you know they would their booth sort of highlighted this all-in-one sort of convers you have your AI agents doing QA you have your

you know back to the point of what I'm seeing about the UI shifting around who they're anticipating the user to be and increasingly the talk around non-technical users. You know, I saw a demo from Vappy, for example. Vappy allows you to build your voice agent. It has a section for simulating conversations, but it's a vibe code product. So, you need to vibe, you need to prompt your expected personas. I I kind of see their thinking, but I don't I don't know how actual like intuitive that is. So, you know, you describe it out and it go it gets to work. So, you no longer have to like do things. And I guess that's what I'm seeing being a trend as well. And I know I I haven't answered your question about which one blew my mind away.

I I think the one that the demo was most in preface is probably Sound Hounds Amelia. I really respect their UI and just the [clears throat] the solution engineers are very uh well-versed and walk you through everything and I I thought that was very impressive. Whereas Sierra perhaps because I was a vendor but even not a competitive vendor I didn't get a real look into their product. Uh but Amelia and Sound they just they really go into detail about uh the ins and out of their product. So yeah that's that's been a bit about my impressions on the expo floor. It's interesting that you mentioned Amelia because that was an acquisition by Soundhound and I was just looking it up.

That was formerly known as IPOF which was founded in 1998 and they you know they've been in this space forever. So kudos to them I guess for for staying up to date and and refreshing the technology. I certainly do have regard for those players that have been around for some time and do know, you know, now that we've worked in the contact center for a number of years, there are a lot of tricks to it that you have to know. It's not to say that there's not room for a lot of new technology, but there's stuff, you know, that we've seen customers really struggle with like alpha numeric codes, recognizing people's names, you know, these sort of classic thorny speech recognition issues that you can't just assume that the new stuff solves. And so there is a part where the new vendors have to kind of learn about the space and and obviously some of the older vendors they need to refresh their stacks. The other thing that what you're talking about there gets me thinking about is like I mean you say Genesis and Sierra that to me is such a great you know contrast between those two. I mean Genesis they are in such a great position in terms of the channel. They are in this dominant position in terms of their market share in the CCA space for so many people. If you want to go live with AI, you're going to have to do it through them. And I don't, you know, they do have an offering, but they haven't moved incredibly quickly to refresh their stack for Gen AI, you know, so they've got the channel, but they don't necessarily have the latest technology yet ready for people to use. And then Sierra, uh, you're talking about them, Diego. I mean, and I now I have not seen a Sierra demo, so I should be careful, but I do kind of assume, you know, they have $800 million. They probably built out some pretty good software, and you know, they they sound like they know what they're doing. So, you know, they've got the newer technology. They've got some of the expertise. They've got this forward deployed engineer model, which to me does make a lot of sense for these implementations.

But then how do they deliver it? I mean, it a lot of it kind of has to go through Genesis. I don't know how, you know, maybe those two booths were talking, maybe they're figuring something out,

but that it's going to be really interesting to see how that plays out because people want this technology and, you know, it's still getting figured out how how exactly it's going to land on people's desktops.

Yeah, that really uh brings me back to remembering uh it was Happy Robot was the one that I forgot, but they their booth laid out, you know, their full stack uh

you know, everything in one. But you're right, the expo floor for that was on one side Amazon Connect and Genesis. On the other side it was Sierra, Finn, ElevenLabs and you know Happy Robot and Soundhound as well. Soundhound Amelia. Another thing worth noting there is uh we started to see them come up a lot and especially on the drive-thru headlines. So, I think that's a bit of a shout out to our next episode on on drive-throughs and the history of, you know, the AI use case there. So, you know, we we've started to see them uh pop up as well.

And I think what you're what you're laid out there is totally true, the the tech and channel and the ones that have the channel, but you know, they're catching up with the tech. So, I thought that was super interesting.

Yeah. I mean, so something I wanted to share that I learned this week was a customer wanted us to help them with an Aentic AI implementation and to get a head start on their testing and some simulations for it, but all they had as artifacts to document it so far are some PNG flowcharts. And you know, when they first mentioned this to us, I thought, well, I don't know if that's going to work or not. you know, I don't know if that's something that we can really leverage as a way to start doing some in-depth testing on this system, but you know, we ran it through well, I ran it through Gemini, another engineer in our team, I think, used Claude, and it did a great job. So, I was impressed. I mean that this, you know, the system was able to essentially take just a a flowchart that's a graphical image and basically turn it into a pretty darn good JSON file that then we could leverage to generate tests. So, it's useful to us.

It's it's one more thing that we can offer to our customers. I mean, a point that they're always concerned about is how do we get started quickly? How do we get started with existing artifacts that we have so that we can sort of jump start our QA? And it's nice that this is now another route that's available. So, I thought that was, you know, it was neat. It was practical and also kind of amazing. I mean, I'm always impressed by what generative AI can do. And this is one more use case that I thought was really pretty neat.

I think that's uh it's really exciting. And in my conversations with prospects, getting started is the I think the the hardest part. And every every person has a different set of baseline. Some people have those artifacts and some people don't. So that is really cool to see how that's being addressed. I think we covered a lot today and I'm very excited to keep the conversation going. So this was our first run at this and we still don't have a podcast name besides the bispoken AI show. So if you have ideas, feel free to comment and please be sure to follow us on LinkedIn. We appreciate you joining the conversation and if you have some things to say about anything we touched on, reach out. Maybe we could have you on. So, uh, John, if you have any parting words.

Well, I guess I mean just on that end, I mean, this is something that, you know, we want this to be timely and of the moment, and part of what's timely or of the moment is us figuring out what we want to do with the show. So, if you have suggestions on things you'd like to see us cover, we do like talking about the news. I think that the developments that are happening in AI, I mean, they're just amazing. And so, certainly part of our inspiration for doing this, I really like the 20 VC show where they talk about the latest developments. It's it's just amazing. It's a really exciting time generally in technology, I think, particularly for our space of voice AI. It's more exciting than I ever thought it would be in the time that I've spent it. So, I it's really cool, you know. So, talking about development, I think is obviously, you know, that's that's something we wanted to do. But then we were planning on doing some things in which we do sort of history or sort of reviewing uh stuff that's happened and some of you know how the technology has come to be.

you know, we find that uh to be sort of compelling and we've got some ideas there as Diego was mentioning talking about the history of uh voice AI and fast food drive-through ordering. Going through some use cases that we see as well as having on guests. So, those are some of our ideas for where we could see this going, but if you have any thoughts, we appreciate them.

Thanks a lot, John, and thank you for listening. Until next time.

Thanks a lot.