Latest / CE Pro Podcast / CE Pro Podcast #174: Sonance CEO Ari Supran and How the Custom Integration Channel is Using AI

Transcript

CE Pro Podcast #174: Sonance CEO Ari Supran and How the Custom Integration Channel is Using AI · CE Pro Podcast

• 0:09
  SPEAKER_00: Everybody, uh welcome to the C Pro podcast.
• 0:12
  SPEAKER_00: Uh this is Zach Ricomo, executive editor, and today I have a special guest, uh Ari Shapron, the uh CEO of Sonance.
• 0:20
  SPEAKER_00: But uh we're not gonna talk about um audio at all.
• 0:22
  SPEAKER_00: But Ari, uh what's going on?
• 0:25
  SPEAKER_01: Hey doing Zach.
• 0:26
  SPEAKER_01: Good day.
• 0:26
  SPEAKER_01: Thanks for having me.
• 0:27
  SPEAKER_00: Of course, of course.
• 0:30
  SPEAKER_00: Um yeah, so so we'll maybe we'll talk about audio, but we're today we're gonna talk about um AI.
• 0:37
  SPEAKER_00: Maybe we should set the stage a little bit about how how we met, because I thought it was kind of funny.
• 0:41
  SPEAKER_00: Um so I as you know, uh spring conference last year.
• 0:46
  SPEAKER_00: Um I don't remember uh what day it was, but I believe might have been game seven of the uh conference.
• 0:52
  SPEAKER_00: Uh I can't remember which round it was now, but uh the Sounds Explain the Knicks.
• 0:56
  SPEAKER_00: I'm a Boston sports fan, like you read about.
• 0:59
  SPEAKER_00: Um and so I'm watching the game at the bar, and there's this guy next to me cheering for the Knicks.
• 1:06
  SPEAKER_00: I'm like, who the hell is this guy?
• 1:08
  SPEAKER_00: Get this guy out of here.
• 1:10
  SPEAKER_00: And then then it's Ari.
• 1:11
  SPEAKER_00: So we we start talking.
• 1:12
  SPEAKER_00: I haven't met him before.
• 1:13
  SPEAKER_00: Uh so he tells me about a presentation that he that he's gonna uh gonna do the next day.
• 1:18
  SPEAKER_00: Um you know, he's like, yeah, like AI and like, you know, in the home, those use cases are are pretty sexy, right?
• 1:24
  SPEAKER_00: Well, that's not what I'm talking about.
• 1:26
  SPEAKER_00: Um so you know, the the presentation and and Ari can speak to this uh far better than I can, but it's really about how integrators are using it or and can use it to uh just you know make themselves more efficient and free up time to do more impactful creative stuff.
• 1:40
  SPEAKER_00: Um so that's what we're talking about.
• 1:42
  SPEAKER_00: Um Ari, how how did you how did this all start for you?
• 1:46
  SPEAKER_00: You know, well when did something click or like I need to start looking at AI and in ways to be more efficient?
• 1:53
  SPEAKER_01: Yeah, I guess it started just um probably, I guess it feels like two years ago or so, pretty early on.
• 2:00
  SPEAKER_01: I was I wasn't like day one of Chat GPT when it first came out, but in the first month or two, and I just started to use it myself, and and you know, I'm not a computer science, I have no computer science background.
• 2:15
  SPEAKER_01: Uh I was a geography sociology double major.
• 2:18
  SPEAKER_01: Um, that has uh served me well.
• 2:20
  SPEAKER_01: And um I just started to use it like for practical things in life and started to then show some friends and family, and and and then what happened was um one of uh one of our sonance dealers out here in Southern California, Audiovisions, Mark Hoffenberg, um, good friend of mine, and that you know, they're right here in Orange County with a few offices around California, and they've got about a hundred employees or so, and they often do their annual all-hand summit in our facility.
• 2:52
  SPEAKER_01: We make it available to them, and he's asked me to uh come back to do a keynote address to his team because I had done one a few years prior.
• 3:02
  SPEAKER_01: Um, and that one was on culture and some of our cultural beliefs here at Sonance, and it was well received.
• 3:07
  SPEAKER_01: And and like I said, Mark and I are tight.
• 3:09
  SPEAKER_01: So he said, uh, well, you know, I think my team's really heard you speak about culture.
• 3:14
  SPEAKER_01: So what else can we talk about?
• 3:15
  SPEAKER_01: And I said, Well, you know, I don't know.
• 3:18
  SPEAKER_01: I gave him a couple ideas, but I said, you know, there's this thing I've been kind of finding enjoyable lately, which is showing people how to find practical uses of AI.
• 3:27
  SPEAKER_01: He says, Oh my gosh, that is the number one thing on the survey we did for internal employees of what you want to learn.
• 3:33
  SPEAKER_01: And this is probably a year and a half ago or so.
• 3:36
  SPEAKER_01: I said, All right, well, I'll uh I'll see, I'll I'll I'll prepare a presentation.
• 3:40
  SPEAKER_01: And so we had that uh here in our our facility in San Clemente, in our studio, and it was it was well received and got a lot of people, you know, kind of kind of um got a lot of people seeing AI a little differently.
• 3:53
  SPEAKER_01: And again, I I'm very clear, I'm like, I'm not the person to get up and talk about machine learning and or or or where AI is going to impact society in the future.
• 4:02
  SPEAKER_01: I have thoughts um for sure, but I'm no expert.
• 4:06
  SPEAKER_01: But my whole lane has been how to how to harness AI for practical purposes to upskill people, and that was well received and kind of snowballed from there.
• 4:16
  SPEAKER_01: Uh, I got a call from Travis Leo, it's the Synergy Group, and asked me to come down to Mexico.
• 4:21
  SPEAKER_01: It would have been a little over a year ago now.
• 4:23
  SPEAKER_01: And um a couple people on the ASEAN board board were at that presentation.
• 4:28
  SPEAKER_01: It was also uh seemingly well received, and and and that got uh the idea up to Patrick at ASEON to have me speak there.
• 4:34
  SPEAKER_01: And and and now I've, I don't know, a few few dozen times uh uh between podcasts and presentations.
• 4:40
  SPEAKER_01: I just I love it.
• 4:41
  SPEAKER_01: It's it's uh there's something about um imparting this, you know, opening people's eyes to something when they have their like wow moments with it, and then they they call you back and and and uh say thank you.
• 4:56
  SPEAKER_01: I mean, there's just I can't think of anything more gratifying than that.
• 5:00
  SPEAKER_01: Right.
• 5:00
  SPEAKER_01: Um so I I know that um it's gotten a little life of its own, and I'm I'm I have a hard time saying no, but I now have a team and we're dividing and and and I've got others uh enlisted to help educate our customers, but quite frankly, it's not something exclusive to Sonyance uh customers.
• 5:19
  SPEAKER_01: It's uh it's just part of our ethos of of uh being an authentic partner and and giving back wherever we can.
• 5:25
  SPEAKER_01: And you you invited me to be on the smart stage at CDIA.
• 5:28
  SPEAKER_01: I really enjoyed that.
• 5:29
  SPEAKER_01: And I did a presentation with Alex Caposlatro from Josh AI, who who truly is a AI uh you know, I'd hope so.
• 5:38
  SPEAKER_01: You know, um expert, but we had a really good uh lot of fun with that.
• 5:42
  SPEAKER_01: And and he he he he did a little bit more on the science and theory, and I I kind of complimented that with a little more practical uses and and we had we had a good it was good.
• 5:53
  SPEAKER_01: So yeah, that's how it started.
• 5:55
  SPEAKER_01: That's how it's been going, and um uh it's been um it's been fun, really.
• 6:01
  SPEAKER_00: Yeah, yeah.
• 6:02
  SPEAKER_00: Um yeah, the presentation, uh the first time I saw it is awesome.
• 6:06
  SPEAKER_00: Um it was you know uh kind of a similar presentation on the smart stage, um, you know, a few months after we first met um at Azion, standing room only.
• 6:16
  SPEAKER_00: Um, and I've only heard good things about other presentations you've done on this topic.
• 6:20
  SPEAKER_00: Um so just just to kind of uh piggyback on something you said, um, you started playing around with it um just for kind of like personal use cases.
• 6:28
  SPEAKER_00: If you can share what were some of those early ones and you're then where like something clicked, you're like, oh, I I need to bring this to my professional life.
• 6:36
  SPEAKER_01: Yeah.
• 6:37
  SPEAKER_01: I mean, I I think you know, my my life and my work are so intertwined.
• 6:41
  SPEAKER_01: I'm trying to remember the very first use cases for it.
• 6:44
  SPEAKER_01: Um, but I I I think it was obviously early days of of chat and just you know, that that being able to summarize something quickly, be able to write an email, that that that basic stuff just I just felt like there was I could already see there was something to it.
• 7:00
  SPEAKER_01: I I remember early days of of trying out Anthropic's uh product Claude and just thinking, okay, there's something even a little more different about this.
• 7:10
  SPEAKER_01: I sort of, I don't know, I feel like I related better to uh to Claude, and I've become a big anthropic user ever since.
• 7:18
  SPEAKER_01: And it's uh proven fruitful as they've really expanded and grown their capability.
• 7:24
  SPEAKER_01: And and obviously, as as everybody's seen in the in the press, or I don't know, maybe I live in a bubble and and people aren't even aware, but they've seemingly overtaken open AI for the moment in uh in revenue.
• 7:33
  SPEAKER_01: And a lot of it is an enterprise, right?
• 7:35
  SPEAKER_01: They they really focused on enterprise and not the consumer.
• 7:38
  SPEAKER_01: And we're a big anthropic shop here.
• 7:40
  SPEAKER_01: I've got almost a hundred employees on an anthropic plan already.
• 7:45
  SPEAKER_00: So that's great.
• 7:46
  SPEAKER_00: Um, I do remember during your your your ASIAN presentation that uh one of the use cases was um you booked dinner and got an Uber for your wife and and your friends.
• 7:56
  SPEAKER_00: Yes, yes, um, all just from um just just a simple prompt.
• 8:00
  SPEAKER_01: Yeah, that was uh that was Perplexity, which is I'm a big fan of Perplexity.
• 8:04
  SPEAKER_01: Um, and uh they have an iPhone app, I'm sure it's available on Android as well, that has a voice uh control or voice um feature to it, and it has these connectors into OpenTable and Uber and different things, and and it's one of to me, one of the the the two crimes of technology are that Sirius is still as bad as it is, and that products like Perplexity are a better version of it.
• 8:29
  SPEAKER_01: And the fact that, you know, and we're a Microsoft shop here at Sonance, but the fact that Anthropic has created a better Microsoft copilot than Microsoft has, although I do hear that it's copilots gotten a lot better recently.
• 8:43
  SPEAKER_01: But there was a time two years ago, year and a half ago, where at Sonance we said, well, we're a Microsoft shop, we need to just use Copilot because it's going to be the best.
• 8:52
  SPEAKER_01: And I I shudder to think had we stuck with that mentality the entire time and not branched out and tried something like Claude, where we would be in our AI adoption.
• 9:02
  SPEAKER_01: And I'm really pleased at it uh, you know, I feel like we took a while to push the boulder up the hill here, but I I think it's crusted and we've got real traction.
• 9:10
  SPEAKER_01: And we're doing some really uh for me, really um impactful things at Sonance to upskill employees and to create more efficiencies.
• 9:20
  SPEAKER_01: And I could go on.
• 9:21
  SPEAKER_01: There, there's so many things that we're doing, but I'm I'm equally impressed with some of the things that integrators that I know are doing with AI.
• 9:28
  SPEAKER_01: And I know you've seen some of it yourself, and it's uh it's exciting.
• 9:32
  SPEAKER_00: Yeah, yeah.
• 9:32
  SPEAKER_00: And a lot of them uh uh credit you.
• 9:35
  SPEAKER_00: Um I mean you've been on uh you've been you've given this presentation dozens of times now.
• 9:40
  SPEAKER_00: Um so you know what's what's your overall message to to folks that are listening to it?
• 9:47
  SPEAKER_01: My overall message is that there is an arbitrage moment of intelligence and capabilities that has never existed before and may never exist again right now.
• 9:58
  SPEAKER_01: And by that I simply mean that there is the vast majority of the people that I present to, when I am presenting to them, unless it's like a secondary advanced class that we've done and things like that, it you know, you do a show of hands and what you find is that everybody's used AI.
• 10:15
  SPEAKER_01: And everybody in the audience, most people in the audience, think of AI as this singular monolithic thing.
• 10:21
  SPEAKER_01: They're like, oh, AI is good at this, it's not good at that.
• 10:23
  SPEAKER_01: But then you dive a little deeper and you realize that they're only using a version of AI that I refer to as chat or is referred to as chat.
• 10:33
  SPEAKER_01: I don't mean chat GPT, although that is a chat, but even Cloud Chat, Gemini chat, and and I just my message is um that we're past chat, right?
• 10:44
  SPEAKER_01: And and chat is a uh a format where you ask a question and your goal is to get an answer.
• 10:49
  SPEAKER_01: And in my opinion, we're past that.
• 10:52
  SPEAKER_01: And tools like Claude Cowork, Claude Code, which are anthropics products, and on the OpenAI side, Codex, which is their coding harness, but can be used for what I do on a daily basis, which I refer to as knowledge work, or is referred to as knowledge work.
• 11:10
  SPEAKER_01: What do I mean?
• 11:11
  SPEAKER_01: Excel spreadsheets and documents and writing and PDFs and slide decks and these these tools like Cloud Co-Work and Codex use a coding harness and you say, Well, I don't program, that's not my job.
• 11:29
  SPEAKER_01: I run a business or whatever.
• 11:30
  SPEAKER_01: And it's just people don't realize is that, and this is why Anthropic grew so fast is that you know what an Excel document is at its at its heart is code, and so is a slide deck.
• 11:40
  SPEAKER_01: And so these coding harnesses, like Codex and Cloud Cowork and Cloud Code, are extremely powerful and capable at doing knowledge work.
• 11:49
  SPEAKER_01: And so back to to what my message is, is that if you're not using these tools, you really don't, I hate to say it, but you really don't know what AI can do and what it can't do.
• 12:02
  SPEAKER_01: And once you do, and you're not gonna listen to me, right?
• 12:05
  SPEAKER_01: I could show you, I could tell you, but everybody needs to have their wow moment with it.
• 12:11
  SPEAKER_01: And it's a personal thing where it exceeded your expectations.
• 12:16
  SPEAKER_01: For me, I I the thing I found the most is especially I and and I teach a lot lately about using Claude Cowork.
• 12:22
  SPEAKER_01: And if any listeners out there have not tried Claude Cowork and are still in chat, I just couldn't encourage you more to do to try Claude Cowork.
• 12:31
  SPEAKER_01: So what the heck's the difference, right?
• 12:32
  SPEAKER_01: Well, the difference is that Claude Cowork is more of an agentic product.
• 12:36
  SPEAKER_01: The goal of every session, every time I sit down with Claude Cowork, my mentality is that we are going to get something done and you're gonna do it for me.
• 12:46
  SPEAKER_01: Um, so I'm I'm in a delegation mode.
• 12:48
  SPEAKER_01: I treat it, for me, it's one of the best named products in all of technology because it truly feels like I am working with a coworker.
• 12:58
  SPEAKER_01: And and not just because I can talk to it like a coworker, but because it acts like a coworker.
• 13:04
  SPEAKER_01: And what do I mean?
• 13:05
  SPEAKER_01: So it I point it to a folder with files in it.
• 13:09
  SPEAKER_01: I'm not uploading a document to it.
• 13:11
  SPEAKER_01: I put my documents in a normal folder.
• 13:14
  SPEAKER_01: I'm on a Mac, so that's in Finder, but on a PC, I guess it's Explorer.
• 13:18
  SPEAKER_01: And I might have 42 documents in it: three Excel spreadsheets, some PDFs, a slide deck, a PDF, numerous things, right?
• 13:26
  SPEAKER_01: I point it to that and it analyzes and reads those documents quickly and then starts to ask me questions about what I'm trying to do.
• 13:36
  SPEAKER_01: We, we, we, it interviews me to get on the same page of the task that I'm asking it to do.
• 13:42
  SPEAKER_01: And that's for me the first moment of like, okay, this is different.
• 13:46
  SPEAKER_01: Because the questions that it asks are so insightful that they're grounded in the information in the files that it saw.
• 13:55
  SPEAKER_01: And that's where it's like, I've seen so many countless people that I have taught how to use cloud cowork.
• 14:02
  SPEAKER_01: It is that moment that they start to realize things are different.
• 14:06
  SPEAKER_01: Because much like an employee, a coworker, that I would hand a folder to and say, familiarize yourself with everything in here, ask me questions to make sure you understand it.
• 14:16
  SPEAKER_01: And then what I'm gonna want to do is build a report, a dashboard.
• 14:19
  SPEAKER_01: And that's the key thing, too, is in a chat interface, you ask things like, What is this?
• 14:25
  SPEAKER_01: Where is that?
• 14:26
  SPEAKER_01: Those are questions that beg an answer.
• 14:28
  SPEAKER_01: In a cowork or codex world, it's words like build me, create this, design that like it's an action, it's a delegation.
• 14:40
  SPEAKER_01: And so when it then asks those questions back about, hey, what I see, you know, and then the thing I always tell people, first thing you should do, point it to your downloads folder, which I'm sure like me is a mess.
• 14:52
  SPEAKER_01: You know, and it looks and you say, help me clean up my downloads folder.
• 14:55
  SPEAKER_01: Now, in a chat interface, it would know nothing about what's in there and start asking you random generic questions.
• 15:02
  SPEAKER_01: But in a co-work session, it looks at that and says, Well, Zach, I see that you've got 74 documents in here.
• 15:09
  SPEAKER_01: It looks like it's a mix of personal and work.
• 15:11
  SPEAKER_01: Why don't I suggest that we 74?
• 15:14
  SPEAKER_01: Yeah, but that's it.
• 15:14
  SPEAKER_00: Those are brookie numbers.
• 15:16
  SPEAKER_01: Okay.
• 15:16
  SPEAKER_01: So it might say, you know, hey, why don't we organize this by work and personal?
• 15:21
  SPEAKER_01: And then inside of those, do you want to organize it by project and then by file type or file type and then by project?
• 15:28
  SPEAKER_01: And then it says, like, oh, in the next question, hey Zach, I see you've got seven versions of the same spreadsheet.
• 15:35
  SPEAKER_01: They're all the same, except this one, you know, this one's the most recent.
• 15:39
  SPEAKER_01: Why don't we delete all of them except for the most current one?
• 15:43
  SPEAKER_01: And you're like, yeah.
• 15:45
  SPEAKER_01: Like, yes, thank you.
• 15:47
  SPEAKER_01: And that's the beginning.
• 15:48
  SPEAKER_01: That is the beginning of this feeling, you know, there's an expression in the AI world of, can you feel the AGI?
• 15:56
  SPEAKER_01: Okay.
• 15:57
  SPEAKER_01: We are clearly not at AGI yet, but we get glimpses of it.
• 16:02
  SPEAKER_00: Yeah.
• 16:02
  SPEAKER_01: For me, Claude Cowork, when it started to ask me these insightful questions, was the beginning of feeling what AGI, what I imagine AGI will be like.
• 16:13
  SPEAKER_00: Yep.
• 16:14
  SPEAKER_00: Yeah, absolutely.
• 16:15
  SPEAKER_00: And and honestly, if you use the the chat functions enough and give it enough information and upload enough documents, it just knows everything about you.
• 16:24
  SPEAKER_01: Um I know people are uncomfortable with.
• 16:28
  SPEAKER_01: Yeah.
• 16:28
  SPEAKER_01: Um as if Google doesn't know everything about us already anyway.
• 16:31
  SPEAKER_00: Yeah, yeah, yeah, absolutely.
• 16:34
  SPEAKER_00: Um yeah, so so you know, you've clearly inspired people uh to begin their AI journey, and then I met I met uh you know a handful of them at HTSA uh a few weeks ago now in Fort Lauderdale.
• 16:46
  SPEAKER_00: Um so and now that the and I saw some really impressive tools that they used, uh that they that they created using these tools.
• 16:53
  SPEAKER_00: Um, and now that they're beginning to kind of like you know understand it, uh what what's next?
• 16:58
  SPEAKER_00: What separates them from the mature users?
• 17:01
  SPEAKER_01: Yeah, it's well I would say this.
• 17:03
  SPEAKER_01: I I th the first thing I would say is I don't know exactly everybody that you talk to, but the ones that I see, whether you know HTSA, AZIO, and ProSource doesn't matter.
• 17:11
  SPEAKER_01: I see examples of this throughout that are building really impressive tools.
• 17:16
  SPEAKER_01: And and it starts with this notion that you know not everybody's aware of or maybe aware of and not bought into, but I know I am, which is that it is harder to purchase a piece of software from a software vendor and wrestle and compromise with it to get it to do what you need in your business than it is to just simply build a piece of software purpose built for your company exactly the way you want it today.
• 17:47
  SPEAKER_01: And yes, I am saying building software using natural language only.
• 17:53
  SPEAKER_01: So you know, as they as the expression goes, the the the the most popular coding language in the future is going to be English.
• 18:01
  SPEAKER_01: And I agree.
• 18:02
  SPEAKER_01: Okay.
• 18:03
  SPEAKER_01: And you know, I always say this on stage too.
• 18:05
  SPEAKER_01: Someone told me probably a year and a half ago that in the future we're going to in the moment create a piece of software, use it once, and discard it.
• 18:17
  SPEAKER_01: I thought they were nuts.
• 18:19
  SPEAKER_01: And I do it all the time now.
• 18:21
  SPEAKER_01: So, what's an example of that?
• 18:22
  SPEAKER_01: I I could be sitting there on my phone, you know, I don't know, I'll watch in the next game and saying, you know, uh, show me all the places Jalen Brunson's, you know, made his shots from in the you know quarterfinals and where he misses from.
• 18:36
  SPEAKER_01: And it'll spin this little dashboard up for me with a basketball court and show me that.
• 18:41
  SPEAKER_01: And I check it out.
• 18:42
  SPEAKER_01: That software is over after the next game.
• 18:45
  SPEAKER_01: It's no longer relevant.
• 18:46
  SPEAKER_01: I'll never use it again.
• 18:47
  SPEAKER_01: Uh I'm a big hockey fan.
• 18:49
  SPEAKER_01: I did some, you know, one I remember doing this that was like wow.
• 18:52
  SPEAKER_01: Me was like, all right, the NHL drafts coming up.
• 18:54
  SPEAKER_01: Show me who the Rangers should pick.
• 18:55
  SPEAKER_01: And it did the research, built this little dashboard, and allowed me to choose is is offense more important than defense, you know, whatever.
• 19:02
  SPEAKER_01: And it, oh, well, then I'd rank them this way for the Rangers.
• 19:06
  SPEAKER_01: And I used it once, discarded it.
• 19:08
  SPEAKER_01: And so making software is becoming easier and easier.
• 19:11
  SPEAKER_01: And I know that there's many software developers out listening that might say, yeah, you don't understand, and it's not okay.
• 19:18
  SPEAKER_01: Um, to each their own.
• 19:20
  SPEAKER_01: All I know is that and to be clear, we build software here at Sonic using AI, in part because we have a technical team backing them up.
• 19:30
  SPEAKER_01: And so we are able to get business users who know nothing about software development to build apps that I wouldn't feel comfortable allowing them to do and use in the business if I didn't have, if I hadn't had made the investment in building out a technical team to back them up to make sure that we're doing it in a scalable, secure, robust way.
• 19:54
  SPEAKER_01: So back to your question, sorry.
• 19:57
  SPEAKER_01: What I see them doing is not just using cowork and chat and using AI in the traditional sense.
• 20:04
  SPEAKER_01: They are building tools.
• 20:05
  SPEAKER_01: And what they've discovered is that in the old world, it made sense that, oh, we need we're a Google shop, everything should be Google, or we're a Microsoft shop, everything should be Microsoft.
• 20:18
  SPEAKER_01: But there is a newfound interoperability of software tools in the AI world that has never been seen before.
• 20:27
  SPEAKER_01: And it is, to me, changing everything.
• 20:30
  SPEAKER_01: And so what do I mean?
• 20:31
  SPEAKER_01: These these things that you are commonly referred to as connectors in AI tools, but technically is what's called a model context protocol or MCP, as well as standard APIs that we're more familiar with, or something called CLIs or command line interface interfaces.
• 20:48
  SPEAKER_01: But all they wallet means is that these are um easy integrations between tools.
• 20:54
  SPEAKER_01: So for instance, we uh use Microsoft Teams, but I also have Slack, obviously two different companies, but I'm able to build myself a dashboard across them both.
• 21:06
  SPEAKER_01: And that just wasn't possible before.
• 21:08
  SPEAKER_01: And I think so what I see integrators doing, um the ones that that I think you're referring to and that are doing some impressive things, is they are building tools to help solve problems and run their business on that I consider to be abstraction layers above some of the SaaS tools that they already have in their business.
• 21:28
  SPEAKER_01: And some of them are saying, I don't actually need that tool anymore.
• 21:32
  SPEAKER_01: I'm just gonna recreate a version of it for myself.
• 21:35
  SPEAKER_01: And that's where I see things going.
• 21:36
  SPEAKER_00: Yeah.
• 21:37
  SPEAKER_00: Yeah.
• 21:38
  SPEAKER_00: Uh should this have some of the uh the the software vendors in uh our space, you know, a little bit on their heels?
• 21:48
  SPEAKER_01: You know, I I I it's a good question.
• 21:50
  SPEAKER_01: I I don't claim to be an expert that that has a good hand on this.
• 21:55
  SPEAKER_01: Here's what I would say.
• 21:56
  SPEAKER_01: I saw that Salesforce um market.
• 22:00
  SPEAKER_01: Benny off put out an announcement about two months ago.
• 22:02
  SPEAKER_01: And what they said was that the future, our future, so this is Salesforce and Slack and everything, our future graphic user interface or user interface is our connector, meaning we recognize that we have to lean into what the moat that we can create of what we do really well, which is not necessarily that everyone should come to our tool and use our graphic user interface.
• 22:32
  SPEAKER_01: And so they're really leaning into this exact idea, which is make sure that our software is extremely well suited to connect to and serve up our data and information to some other abstraction layer above it.
• 22:48
  SPEAKER_01: For us at Sonance, that's Microsoft Teams.
• 22:51
  SPEAKER_01: Okay, so we use Microsoft Teams for the normal things that Microsoft Teams is used for.
• 22:56
  SPEAKER_01: But all of our internal apps that we're building throughout our organization, they will all appear to our employees as apps inside of Microsoft Teams.
• 23:06
  SPEAKER_01: What we learned, or what I learned, maybe my team knew, is that it's very hard to get your app put onto the Microsoft Teams platform for the world to see.
• 23:16
  SPEAKER_01: It's extremely easy to get it put into just a single tenant, only your employees can see in your Microsoft Teams.
• 23:23
  SPEAKER_01: And so for us, all and we we haven't talked about agents yet, but when we get to a place where we deploy agents as coworkers into our workforce, they will also appear to our employees inside of Microsoft Teams.
• 23:35
  SPEAKER_01: They will DM them like you and I might DM together.
• 23:38
  SPEAKER_01: And so point being is that these other SaaS tools, if they to me, if they're not connected, they're dead.
• 23:50
  SPEAKER_01: That's simple.
• 23:50
  SPEAKER_01: That's it.
• 23:51
  SPEAKER_01: I can't be any more, you know, I can't sugarcoat that.
• 23:55
  SPEAKER_01: That's my opinion.
• 23:56
  SPEAKER_00: Yeah.
• 23:56
  SPEAKER_01: So having a modern open API is the beginning of it.
• 24:00
  SPEAKER_01: Turning that API into a MCP is the next step, but I think that's the future.
• 24:06
  SPEAKER_01: And I think there's going to be a whole newfound revolution of not only app building within businesses, which is exciting, um, but also SaaS tools that are going to make sure that their connector is robust and put more focus on that than they do their own GUI.
• 24:29
  SPEAKER_01: So that people can build these tools as abstraction layers over them.
• 24:33
  SPEAKER_00: Yeah, yeah.
• 24:34
  SPEAKER_00: I mean, to your point, all the tools that that, you know, uh integrators are showing off to me, you know, rely on um, you know, open APIs and and um integrations.
• 24:44
  SPEAKER_00: Um so so let's talk about agents.
• 24:47
  SPEAKER_00: Um uh you know, for for for folks that still think AI is mostly just you know asking Chat GPT questions, you know, oh no, what's the difference?
• 24:55
  SPEAKER_00: Everyone's heard the term agentic AI, but what's it mean in practice?
• 24:58
  SPEAKER_01: Well, I would say to start off with that um agentic AI as a general term, I think that describes what I was saying about cloud cowork and codecs, where you we, you know, to me, chat is generative AI, right?
• 25:13
  SPEAKER_01: We're generating an answer.
• 25:15
  SPEAKER_01: And cowork and codex are more agentic AI where we're delegating a task and it's working agentically to get that task done.
• 25:25
  SPEAKER_01: Agents, as um of the last you know, three or four months when um product, an open source product called OpenClaw, well, I guess it came out uh originally calling itself ClaudeBot, and then had to change its name to OpenClaw, really set the again.
• 25:44
  SPEAKER_01: My little bubble of AI world, I don't want to say the world so many people don't know exists, but if listeners um have that, you know, they've certainly heard of OpenClaw if they're into this, into this uh game.
• 25:56
  SPEAKER_01: It's um what is OpenClaw?
• 25:57
  SPEAKER_01: OpenClaw is and there, and to me, OpenClaw is an idea.
• 26:00
  SPEAKER_01: There's already many competitors to it.
• 26:02
  SPEAKER_01: One called Hermes Agent that I've been using lately that's quite impressive.
• 26:06
  SPEAKER_01: Perplexity has their own version of this called Perplexity Personal Computer.
• 26:09
  SPEAKER_01: Claude is clearly, Anthropic is clearly building their own.
• 26:13
  SPEAKER_01: They have all the building blocks of it, they just haven't put it all together and said, here's our agent.
• 26:17
  SPEAKER_01: But what is this thing?
• 26:19
  SPEAKER_01: It's an agent that runs on a typically a dedicated computer.
• 26:26
  SPEAKER_01: So, like I have a Mac mini at home with no monitor, no keyboard that my agent runs on 247, 365.
• 26:34
  SPEAKER_01: Why?
• 26:34
  SPEAKER_01: Just to separate it from my main computer from a security standpoint.
• 26:39
  SPEAKER_01: But it has uh a couple things that are interesting.
• 26:42
  SPEAKER_01: It has persistent memory.
• 26:44
  SPEAKER_01: So everything I ask it to do, it, you know, and I'm like, no, no, no, uh, this you don't understand.
• 26:51
  SPEAKER_01: Uh iPord is a division of Sonance.
• 26:53
  SPEAKER_01: I'm just making something up.
• 26:54
  SPEAKER_01: It's like, I got it.
• 26:55
  SPEAKER_01: And it writes itself a note and it will remember that.
• 26:57
  SPEAKER_01: And so it learns over time.
• 26:59
  SPEAKER_01: It can run agentically 24-7, it can do things like uh scheduled tasks.
• 27:06
  SPEAKER_01: So I can have it just wake up in the middle of the night and check to see if there's any new tasks and run and do things.
• 27:12
  SPEAKER_01: And and ultimately, the best way for me to describe what these types of agents are is that um once you learn how to use something like Cloud Cowork and have it connect to your emails and and calendar and do things for you, or use it to uh code applications.
• 27:37
  SPEAKER_01: What a agent does is it essentially replaces you at the keyboard and allows you to move up a layer and communicate with the agent using essentially text messages.
• 27:50
  SPEAKER_01: People often use something like Telegram or Discord for a little bit more of a secure than something like iMessage, but you can use iMessage or Teams.
• 27:58
  SPEAKER_01: Like I said, for us, these agents in the future will appear in Teams and we'll communicate with them through DMs and Teams.
• 28:05
  SPEAKER_01: But what you're doing essentially is replacing yourself at the keyboard, telling it what you want to do, and then it works agentically to do it.
• 28:13
  SPEAKER_01: So all you need is, you know, if I've got one bar on my phone and I can get a message off, it's running on a Mac mini at my home with high-speed internet, and we'll go do that task.
• 28:24
  SPEAKER_01: If it takes 10 minutes, an hour, or all night, it'll it'll get it done, and then tell me back, you know, hey, I'm done, and it'll put files in my folders, email me, whatever I ask it to do.
• 28:34
  SPEAKER_01: And that's exciting and scary at the same time.
• 28:39
  SPEAKER_01: I fully recognize that we're not um, you know, we're treading slowly.
• 28:43
  SPEAKER_01: And again, I got a technical team that we've made the investment in that is experimenting with this, and you know, uh, it's an open source product.
• 28:51
  SPEAKER_01: Both OpenClaw and Hermes Agent are open source.
• 28:54
  SPEAKER_01: And um, my team is um building our own version of it, using some of that open source code and taking the best of both of them.
• 29:02
  SPEAKER_01: And time will tell if if this is uh work.
• 29:05
  SPEAKER_01: And there's clearly a a human change management element of this and making sure that we we roll this out in the in the proper way.
• 29:13
  SPEAKER_01: And you know, I can't speak for other companies.
• 29:15
  SPEAKER_01: I I just know what our intentions are, which are to upskill our employees and um and and and and take away some of the menial tasks to free up their time to do more value-added activities that help the customers and help the company.
• 29:28
  SPEAKER_01: It's that simple.
• 29:30
  SPEAKER_00: Can you talk a little bit about the importance of um I know you guys hired you know someone to to you know just solely focus on these AI initiatives?
• 29:39
  SPEAKER_00: Um, but I think so.
• 29:41
  SPEAKER_00: When you talk about like coding and and programming, um, programmers it's probably the the top use case for uh these AI systems we're talking about.
• 29:51
  SPEAKER_00: But there's still programmers creating programs, still educated and trained programmers.
• 29:58
  SPEAKER_00: Yes, it's still them using AI tools to create programs, uh you know, commercial.
• 30:02
  SPEAKER_01: Yes and no, both.
• 30:04
  SPEAKER_01: We have both.
• 30:04
  SPEAKER_01: We have software, pure software developers that are, you know, their whole career has been software development that are now using AI to do all of the coding, but they are not doing it in a as it's called a YOLO mode, right?
• 30:19
  SPEAKER_01: It's just like go and and and no, they're watching everything it's doing, steering it, correcting it when necessary, and are in the details of it.
• 30:27
  SPEAKER_01: But we also have pure business people that know nothing about software development that are building tools.
• 30:35
  SPEAKER_01: Now, what we do is we have a technical person set it up for them, get it started, and then allow to the point that now, you know, that someone's gonna apply their audio acoustic knowledge to build a speaker calculator tool to lay out a system with speakers and and and polar plots and things like that, right?
• 30:56
  SPEAKER_01: Um, and that that combination is super powerful, right?
• 30:59
  SPEAKER_01: I can I can pair up a an AI focused software um developer to back them up, but really unleash the domain expert to build a tool to solve a problem or to make the customer experience better.
• 31:18
  SPEAKER_01: And I think that's really exciting right now.
• 31:20
  SPEAKER_01: And and and I see so much potential there as these tools get easier and easier and better and better to change a culture from that some department says they need a tool and they put in a ticket with IT and say, I could use a piece of software to do the following.
• 31:38
  SPEAKER_01: And then this the IT department either says, All right, well, we'll put that in the queue, or wow, this is a little bit more than we should take on.
• 31:47
  SPEAKER_01: We've got this third-party software development company, we could get a quote from them, nine months,$250,000.
• 31:54
  SPEAKER_01: Those types of pieces of software we are now building inside of Sonance in a month.
• 32:01
  SPEAKER_01: With like I said, with a business person building it.
• 32:05
  SPEAKER_00: But I mean like commercially, um, something you would where you would sell, um, or if you're you know, have folks pay your subscription to, it's still in large part trained software developers doing that stuff.
• 32:18
  SPEAKER_00: Um, but I think for the most part, yeah.
• 32:20
  SPEAKER_00: I don't know what for now, anyway.
• 32:22
  SPEAKER_00: Yeah, yeah, for now.
• 32:25
  SPEAKER_01: I I I get the the the the software developers, developers mentality of, you know, if you're putting a piece of software out that is a SaaS tool that you expect, you know, hundreds of thousands, millions of people to use, that that's different than an internal app that 30 people are gonna use in a factory to track this or do that.
• 32:45
  SPEAKER_01: Um, and not all internal apps are created equally, and and and that's something we're we're uh trying to delineate now.
• 32:51
  SPEAKER_01: And and we've got some, you know, I we call them orchestrators here.
• 32:55
  SPEAKER_01: So an orchestrator is somebody that has a day job at Sonance running a department, or maybe they're not even running an apartment, they're just working in the department, but they've got a proclivity for this, they've got an interest in it, and we've set them up with some back-end uh tools that we've curated and chosen for them in a consistent way so that everybody's doing things the same way, um, and and then set them up to build a tool to solve a problem.
• 33:23
  SPEAKER_01: And then we've got a you know, a little bit of a beta testing program and things like that.
• 33:27
  SPEAKER_01: And then we've got other people who have raised their hand and said, I could, I really want to do this full time.
• 33:33
  SPEAKER_01: And we're setting them up with like, all right, you're gonna take these four departments of the company and you're their, let's call it a super orchestrator for the moment, where your job is to, and I gotta be super clear about this.
• 33:46
  SPEAKER_01: The job isn't to build the tool, the job is to understand the business, understand the problems, talk to the people, figure out what problem is worth solving, build the tool to solve it, and then stay focused on making sure that people learn how to use the tool, that they're using the tool, and the success matrix metric, sorry, is the business outcome.
• 34:08
  SPEAKER_01: And they have to stay focused on that.
• 34:10
  SPEAKER_01: It can't be I built the tool and I don't know if anybody's using it.
• 34:13
  SPEAKER_01: That doesn't work.
• 34:15
  SPEAKER_00: Um talk about uh you know the hype around um new AI models.
• 34:19
  SPEAKER_00: I mean, yeah, all these companies love to obviously market um you know new stuff they're releasing.
• 34:25
  SPEAKER_00: Um but as far as we're concerned, you know, do we need to you know uh use all the new the the these new flashy things or or or is what we have now good enough?
• 34:39
  SPEAKER_01: Well, the first thing I'd say is um I don't know that there's again people that aren't as obsessed as I am and and tracking all this stuff, right?
• 34:48
  SPEAKER_01: And I am, it's it's it's it's it's nights, weekends, it's it's um I love it, right?
• 34:54
  SPEAKER_01: But I think what anyone who's really trying to track this can see clear as day is that we are not slowing down, we're accelerating.
• 35:04
  SPEAKER_01: And the new models aren't coming out once a quarter anymore, not even once a month.
• 35:10
  SPEAKER_01: We're about, in my opinion, to reach them maybe once a week, once every other week, and there's a significant leap.
• 35:16
  SPEAKER_01: And I think this week we're gonna see some uh some major launches.
• 35:20
  SPEAKER_01: Um, however, I don't get super excited about the next most intelligent model.
• 35:26
  SPEAKER_01: Why?
• 35:27
  SPEAKER_01: My personal opinion is that these models are already, again, the right models used in the right way.
• 35:35
  SPEAKER_01: Okay, and that by that I mean like let's take anthropic.
• 35:38
  SPEAKER_01: Their top model is called Opus 4.7 right now.
• 35:42
  SPEAKER_01: Um, and then the next one is Sonnet 4.6.
• 35:45
  SPEAKER_01: Um when you use chat, you just chat with that model.
• 35:49
  SPEAKER_01: But when you use cowork, you're using what's called a coding harness.
• 35:53
  SPEAKER_01: And essentially what that means is that they have taken one instance of OPUS and trained it to use a second instance of OPUS to do coding.
• 36:03
  SPEAKER_01: And that coding can be used to build spreadsheets and slide decks and do knowledge work.
• 36:09
  SPEAKER_01: And when you're using the right models in the right way, I personally believe that they are already intelligent enough to do any of the kind of knowledge work that anybody Epsonians would need to do.
• 36:21
  SPEAKER_01: Now, they're gonna continue to get better.
• 36:24
  SPEAKER_01: They are quite expensive to use.
• 36:26
  SPEAKER_01: You say, Oh, isn't it just a subscription?
• 36:28
  SPEAKER_01: Yes, there's rate limits, and we've got um, you know, Anthropic recently increased the rate limits because we were definitely bumping up against that as a company.
• 36:37
  SPEAKER_01: Um and uh, but these models are already smart.
• 36:41
  SPEAKER_01: So I'm excited for the current level of intelligence to come down in price so that it is more scalable.
• 36:48
  SPEAKER_01: Um, and I personally believe that it is not more intelligent models that companies like custom installers and Sonance need in order to get more out of AI.
• 36:58
  SPEAKER_01: It's something else.
• 36:59
  SPEAKER_01: It's uh let's start with just uh education and understanding of what's possible, right?
• 37:06
  SPEAKER_01: There's that.
• 37:07
  SPEAKER_01: But I think in specifically, it's getting two things right.
• 37:11
  SPEAKER_01: One is data and making sure that the data is clean, organized, accessible, and intuitive to know what data you're connecting to.
• 37:23
  SPEAKER_01: Okay.
• 37:24
  SPEAKER_01: We're doing a lot at Sonance on that front, building out what's referred to as a data lake house.
• 37:29
  SPEAKER_01: We've chosen Databricks as the provider of that.
• 37:33
  SPEAKER_01: And it is an AI native database of all of our data, both structured and unstructured data that uses AI inside of the data lake house to organize, tag, clean, and sometimes transform the data, ready for us to access it with business intelligence tools and visualization tools and other things.
• 37:53
  SPEAKER_01: Our apps will pull from that, et cetera.
• 37:55
  SPEAKER_01: But there's another thing that people I think don't really think about yet, which is knowledge.
• 38:03
  SPEAKER_01: All of the tribal knowledge that's in people's heads in a business, all of the knowledge that's in Slack channels and Teams channels and things like that.
• 38:13
  SPEAKER_01: But in particular, all of the information that is in PDFs and PowerPoints and Word documents and all of that, and Excel spreadsheets that is sitting in a Google Drive or a SharePoint, I have come to learn that I need to think about all of those documents as dead documents.
• 38:32
  SPEAKER_01: What does that mean?
• 38:33
  SPEAKER_01: It means that a dead document is a document that is only ever read or accessed when a human being decides to navigate their mouse to that folder, open up the document, and read it.
• 38:44
  SPEAKER_01: And what we need to do is turn all of those things into live documents.
• 38:48
  SPEAKER_01: And a live document is a document that is typically we think of it as a knowledge graph or a web of information in a WikiLinks link style that one document might say CE pro and then say Zach Como, and that's a hyperlink to a different document that's all about you.
• 39:08
  SPEAKER_01: Okay.
• 39:08
  SPEAKER_01: And then in that one, it might have a hyperlink to something else.
• 39:11
  SPEAKER_01: Okay.
• 39:12
  SPEAKER_01: Um that uh is so why?
• 39:16
  SPEAKER_01: Because then we can have our agents, whether they be a clawed co-work with a human being sitting at the computer or an open claw or Hermes agent, connect to that.
• 39:27
  SPEAKER_01: And then when we ask it to do something for us, it automatically says, let me go grab some context from the knowledge graph that will help me.
• 39:35
  SPEAKER_01: And it infuses that context into its understanding of what we're asking for and into the output that it gives us, making it so much more relevant and helpful to the business.
• 39:47
  SPEAKER_01: And so we're building a knowledge graph out right now here at Sonance.
• 39:51
  SPEAKER_01: Um, for a smaller uh company or an individual, the product that seems to be um caught gone a little viral here is a product called Obsidian with an O.
• 40:02
  SPEAKER_01: It's a free product.
• 40:03
  SPEAKER_01: And technically, Obsidian isn't the knowledge graph itself, it's just a visualizer of the knowledge graph.
• 40:09
  SPEAKER_01: I can share my screen and show you what that looks like if you want.
• 40:12
  SPEAKER_01: But um they oh sure, okay, let's do that.
• 40:14
  SPEAKER_01: Um put this here.
• 40:17
  SPEAKER_01: So this is an example of a knowledge graph here.
• 40:21
  SPEAKER_01: Um and so this organic looking uh diagram here, if I zoomed in on it, is just a series of nodes.
• 40:34
  SPEAKER_01: Each node is a person or a topic, and everything is then connected with these lines that show relationships between this information.
• 40:46
  SPEAKER_01: Okay.
• 40:47
  SPEAKER_01: Um, and I can actually drill into any one of these things and see what's inside of it, but I don't use this by walking up to it and and and searching it as a human.
• 40:56
  SPEAKER_01: It is connected to my Claude instance.
• 40:59
  SPEAKER_01: And every time I use Claude, it goes here and grabs information.
• 41:03
  SPEAKER_01: And we're still building this out.
• 41:04
  SPEAKER_01: This is sort of my personal one as an experiment, um, but we're building one out for the company uh right now.
• 41:10
  SPEAKER_01: And of course, we're figuring out the permission layers.
• 41:13
  SPEAKER_01: There'll be confidential information in it, there'll be information we don't want put in there, let's say compensation information.
• 41:19
  SPEAKER_01: But the vision, and at this point for us, it is it is just a vision.
• 41:24
  SPEAKER_01: We're we I don't want to overstate that we've we've done this and it and it works, is that we can create this living organism that is mining information from meeting transcripts and other things to constantly be um the most aware of what's going on and and all that nuanced information.
• 41:42
  SPEAKER_01: Our business is quite diverse with different channels and different geographies and different departments.
• 41:48
  SPEAKER_01: And it's come, it's come, I've come to realize that no one, including me, at this company of about 360 employees, actually knows what our strategy and what our plans are in every single area of the business.
• 42:05
  SPEAKER_01: Now, my leadership team obviously does, and I've got other executives that know parts of it, but I got too many people that have so much talent and so much heart that just don't understand enough to be really, really helpful.
• 42:20
  SPEAKER_01: And it's not their fault.
• 42:21
  SPEAKER_01: But the notion of actually spending enough time with 30, 40, 50 people to learn what's accumulated over my 20 years in this business and the rest of my executive team that are now together for 15 years is almost impossible.
• 42:37
  SPEAKER_01: And so I'm really excited about the next phase for us of building out a knowledge graph, cleaning up our data, getting it into a data lake house, and connecting everything we do to those sources of data and information and combining them into really insightful help and building things.
• 42:58
  SPEAKER_01: And, you know, as an example, uh, I had one of our employees, or I should say I had, he on his own saw a problem and built a tool.
• 43:06
  SPEAKER_01: And the tool was essentially when one of our reps says, Hey, I've got a dealer I want to put on with Sonance, our current process isn't a good one.
• 43:17
  SPEAKER_01: It's a series of emails with docu-signs that go into spam and not really having visibility for the dealer or the rep of where do I stand in this process?
• 43:27
  SPEAKER_01: The process takes too long.
• 43:29
  SPEAKER_01: It's just not great.
• 43:31
  SPEAKER_01: And rather than just automating a broken process with these automation tools like Zapier and Make.com and N8N, he did the right thing, which is he reimagined the process and he built an app that the rep and dealer get a link to.
• 43:48
  SPEAKER_01: They go to this app, they fill out the information and does some really cool things.
• 43:52
  SPEAKER_01: Like I didn't realize this, but apparently we get an address from a dealer, and then a month later we go to ship them their first order, and that address is not.
• 44:00
  SPEAKER_01: Technically a shippable address, and we have to get the product back from UPS.
• 44:04
  SPEAKER_01: Well, now when they enter their address in, it automatically sends it out to some API with a service for an API call that checks to see if it was a legitimate shipping address and corrects it if necessary, saving countless time down the line.
• 44:17
  SPEAKER_01: Another one, when they put in their credit information, it automatically goes out to a secure API to do a credit check.
• 44:24
  SPEAKER_01: If that credit check is above a certain threshold, it's auto-approved and moved to the next step.
• 44:29
  SPEAKER_01: If it's not, then it goes to my credit department and they do credit checks and different things like that.
• 44:34
  SPEAKER_01: So I bring that up just to simply say when we built it, it my team saw it and they were like, this is amazing, but you don't understand in Europe we do things a little differently.
• 44:45
  SPEAKER_01: We don't actually have dealers, it's through master distributors and in our pro channel, it's different and all these nuanced things.
• 44:52
  SPEAKER_01: And it occurred to me that Claude Code, which is what the tool he built it with, didn't know all those nuances.
• 45:00
  SPEAKER_01: But nor did the employee that was building the tool.
• 45:03
  SPEAKER_01: And if I can insert a knowledge graph in the middle of that process where Claude Code is connected to it, I'm confident that the output is going to be much closer to the bullseye.
• 45:16
  SPEAKER_01: And as this knowledge graph builds over time, I just think it's going to transform our ability to upskill and upknowledge our employees to be able to really understand better.
• 45:30
  SPEAKER_01: You know, I have employees that have the courage to come up and ask me a question, like, hey, are you how do we sell to architects?
• 45:36
  SPEAKER_01: And I'm like, well, no, we don't sell to them, but they're quite important and this is how it works.
• 45:40
  SPEAKER_01: And but I recognize that most people don't have that courage and new employees to come.
• 45:44
  SPEAKER_01: And now a knowledge graph can infuse all that knowledge and all that nuance into everything and allow employees in the onboarding process to just on their own time just query it and learn.
• 45:55
  SPEAKER_01: But I want to be clear, it's not just about a chat bot to ask questions.
• 45:59
  SPEAKER_01: That is not the main benefit.
• 46:00
  SPEAKER_01: The main benefit is that it agentically connects to the tools that we use and infuses that knowledge into everything we do automagically.
• 46:11
  SPEAKER_00: How do you get folks to to buy in on AI?
• 46:15
  SPEAKER_00: Um, I think I think it scares a lot of people for right and wrong reasons.
• 46:20
  SPEAKER_00: Uh so how do you do that?
• 46:22
  SPEAKER_00: You know, training, do you do you is it maybe part of the hiring process at Sonance?
• 46:27
  SPEAKER_01: Do you know what's that like?
• 46:30
  SPEAKER_01: I look, I I I feel that at Sonance we have really hit our stride of traction and adoption.
• 46:38
  SPEAKER_01: That being said, are there people at Sonance that are afraid of this?
• 46:42
  SPEAKER_01: I am sure.
• 46:45
  SPEAKER_01: Overall, though, the vast majority I think are excited and see the opportunity.
• 46:51
  SPEAKER_01: Now, let's be clear.
• 46:53
  SPEAKER_01: A couple things they need to say.
• 46:54
  SPEAKER_01: Number one, we don't have some call center of a thousand people doing the same thing that some massive conglomerate might have.
• 47:02
  SPEAKER_01: That I just it's not my world.
• 47:05
  SPEAKER_01: I don't know what how the what their mentality is.
• 47:07
  SPEAKER_01: Another thing, we're a privately held company.
• 47:10
  SPEAKER_01: We work for two individuals that own the company, and really one are our primary founder, Scott.
• 47:16
  SPEAKER_01: And and people are my employees know him.
• 47:18
  SPEAKER_01: They know his values, they know how much he cares about people and cares about the world.
• 47:22
  SPEAKER_01: He's he's pretty much a full-time philanthropist at this point.
• 47:24
  SPEAKER_01: They know our leadership team.
• 47:26
  SPEAKER_01: Okay, they they they know us as humans, as people, they know our values.
• 47:30
  SPEAKER_01: And then quite frankly, we've built trust long before AI was a term that was used in the business by our actions and how we treat our customers and how we treat our people.
• 47:40
  SPEAKER_01: So when I stand up in front of the company and I say, we're not doing this to eliminate jobs, we're doing this to upskill you and to create more time for you to do more value-added things for the customers and the company, I'd like to think that they trust me.
• 47:54
  SPEAKER_01: Um, and I think they do.
• 47:56
  SPEAKER_01: Now, again, big company, or at least by my standards, a big company, and I'm sure there's pockets that are are scared and afraid, and and and I'm um and I care about that.
• 48:08
  SPEAKER_01: But generally speaking, I think that our employees, and then and then for for the other big one is we've brought them a tool in Cloud Cowork that they've seen, they've had their wow moments.
• 48:22
  SPEAKER_01: We're we're about to do some more workshops and training on it, making sure people are using it and know how to use it.
• 48:28
  SPEAKER_01: We've done a lot of work, you know, we we talked earlier about connectors.
• 48:33
  SPEAKER_01: One of the things our dedicated AI technical team has done that I'm excited about is we've taken all of the tools that we use in our business.
• 48:42
  SPEAKER_01: So Microsoft Teams, um, which is and Microsoft 365, right?
• 48:46
  SPEAKER_01: So Outlook and Teams and everything else in there, and Asana, which is our task management tool, and Salesforce and Power BI, and um also some things for for building apps like GitHub and Superbase, which is our database we use, and Vercell, which is how we deploy apps, and we put them all into a basket that we call the Cortex.
• 49:10
  SPEAKER_01: And then our users only have to connect to one connector, and that gives them access to all of those tools, but on a permission basis.
• 49:19
  SPEAKER_01: So not everybody needs Power BI or those things.
• 49:22
  SPEAKER_01: But it's not just a convenience factor.
• 49:24
  SPEAKER_01: Inside of the Cortex, we've also put some skills and rules that are guardrails telling the model without the employee needing to understand how we want, let's let's say they're building an app and they need a database.
• 49:40
  SPEAKER_01: These employees don't know, like I don't know how to create a SQL database.
• 49:44
  SPEAKER_01: They don't need to know.
• 49:45
  SPEAKER_01: Claude knows how to use Superbase to do that, but we want it done a very specific way so we can be consistent.
• 49:51
  SPEAKER_01: And that's a rule that we've put into the Cortex.
• 49:54
  SPEAKER_01: So when anybody says the app says, yeah, I need to set spin up a database, it reads this rule that we've put in there about how we want it done.
• 50:02
  SPEAKER_01: And other things that we've done, let me give you an example.
• 50:05
  SPEAKER_01: We've taken our brand guidelines and turned it from a PDF into essentially an internal website.
• 50:14
  SPEAKER_01: And we've also included what is called a component library, which is what do I want a drop-down menu or a switch to look like in an app?
• 50:23
  SPEAKER_01: And we built this all into a web-based internal app that is part of this cortex of connectors, so that anybody sitting down with Claude to either, it might just be a dashboard that they build, which Claude refers to as an artifact, or a full stack application using GitHub and South Basin Vercell, it automatically infuses our brand guidelines, our logos, our colors, our fonts, and our component library of switches and drop-down menus so that all the apps we build have a consistent look and feel.
• 50:56
  SPEAKER_01: And then when we put them in teams, it all just feels cohesive.
• 51:00
  SPEAKER_01: So doing those types of things, I think, are um important.
• 51:03
  SPEAKER_01: You asked me, I guess I got to start on this question about um about fear and trust and and and um to to to go back to that topic.
• 51:12
  SPEAKER_01: Again, I just think that you can't I don't believe that you can take a culture that doesn't trust leadership and get them to trust you about AI.
• 51:22
  SPEAKER_01: Um, I think it starts with a culture that does trust, that takes time and consistency and consistent leadership and not having a new private equity firm that just bought in or a new this or a conglomerate.
• 51:38
  SPEAKER_01: Again, I I don't want to speak for other manufacturers and other companies out there, but speaking to these custom installers that own this business is if you've created a culture of trust, then you've got the groundwork to get them to trust you on AI.
• 51:53
  SPEAKER_01: And if you haven't, it's gonna be an uphill battle.
• 51:55
  SPEAKER_00: Right.
• 51:56
  SPEAKER_00: Uh coming up on time, so I just want to end with a couple questions.
• 52:00
  SPEAKER_00: Um, you know, how how has it fundamentally changed the way you guys work at Sonance?
• 52:05
  SPEAKER_00: You know, can you tell us any measurable outcomes?
• 52:07
  SPEAKER_00: And then um this is a little bit of a career ball for you, um, since we're talking about, you know, um these these using these models on the back end, um, but our industry is what we call the smart home industry.
• 52:18
  SPEAKER_00: Um there's there's some level of intelligence um in the home.
• 52:23
  SPEAKER_00: And currently, um, you know, compared I think I think the the really solid applications are on the back end.
• 52:32
  SPEAKER_00: Um what can we look forward to for you know this level of AI in the home?
• 52:39
  SPEAKER_00: Where do you see that going?
• 52:42
  SPEAKER_01: All right, let me take one at a time here.
• 52:44
  SPEAKER_01: So uh at Sonance, and let me be super clear.
• 52:46
  SPEAKER_01: We're on a journey at Sonance.
• 52:48
  SPEAKER_01: We've got tons of traction, we've got some real uh meaningful, impactful things that we've done that are saving people time and making a better customer experience.
• 52:58
  SPEAKER_01: But we've got a lot of things that I would say are in the works that I would be um be reckless of me to simply claim success yet.
• 53:07
  SPEAKER_01: Um, but we I think most importantly, we've built a team, we have a vision, we have an employee base that has seen the light of how this can help them.
• 53:19
  SPEAKER_01: There's trust, and we're building structure, and we've got a team of, like I said these earlier, these orchestrators that have a day job.
• 53:27
  SPEAKER_01: I think it's about 15 to 20 employees now, more that really should be part of that.
• 53:32
  SPEAKER_01: We're identifying some super users that can kind of be out put out into the embedded in Teams to go build tools.
• 53:40
  SPEAKER_01: And we're definitely seeing, you know, in our in our factory up in Nevada where where we make James products, um, we've got an individual up there just building all sorts of tools that really create a better workflow and a better visibility for the process and not only for our employees, but for customers to know where's my custom order and so many things going on there.
• 54:00
  SPEAKER_01: And again, I don't want to claim victory yet.
• 54:02
  SPEAKER_01: Um, but but lots of places where we're uh simply solving problems, creating efficiencies, better visibility, and um and and and I think that our employees are seeing that and they uh and they're excited about it.
• 54:19
  SPEAKER_01: And and and and I'll let me throw this in there as well.
• 54:22
  SPEAKER_01: I I get this a lot.
• 54:23
  SPEAKER_01: Like, why are you the CEO?
• 54:26
  SPEAKER_01: Shouldn't you be delegating this and not be so hands-on with the education or or or use of it?
• 54:33
  SPEAKER_01: And I have the same answer because I've thought long and hard about this.
• 54:37
  SPEAKER_01: Um, the first thing I my advice to these smaller, any business, I it doesn't have to be the CEO or the founder, but some leader, to me, you can't fully delegate this.
• 54:48
  SPEAKER_01: Some leader has to have their own wow moment, then get the rest of a leadership team to have some semblance of wow moment, to have the courage to make the investments that I think are required, not just in subscription, but maybe getting either an existing employee or a dedicated hire that's technical enough to be able to backstop this.
• 55:09
  SPEAKER_01: For us, we hired.
• 55:10
  SPEAKER_01: We also took a leader of our business from our iPort business, 15 years with a company, incredibly visionary person to move into a dedicated role to lead this for us.
• 55:21
  SPEAKER_01: So we've made and then hired some other people, moved some people from other departments.
• 55:26
  SPEAKER_01: We've made these investments, but it didn't come by delegating this to the IT department.
• 55:31
  SPEAKER_01: In fact, I think that's in some ways counterproductive.
• 55:33
  SPEAKER_01: I I I have uh I see it where people, you know, just anything that's AI related, they send, you know, check this out, but they don't even understand it themselves.
• 55:43
  SPEAKER_01: Um so I I think that that's all important um there.
• 55:48
  SPEAKER_01: And then to finish your last question, sorry, was about okay, AI in the home.
• 55:52
  SPEAKER_01: Look, I I I'm I'm I see a lot of potential and I see things like what Josh AI is doing, leveraging AI in the home, and it's exciting.
• 56:02
  SPEAKER_01: Um, I've seen some integrators working on some tools that embed um a chat interface for the client when something's broken and how to fix it, and that's tied into to something like Domots or something to reboot things, and I think that's exciting.
• 56:20
  SPEAKER_01: And I think we're literally just scratching the surface of this.
• 56:24
  SPEAKER_01: Um, but uh I think there's a world of opportunity, and uh, I I've I again I don't want to disclose confidential information, but I I have customers and friends that are building some really cool things that are client-facing in-the-home um add-ons to the traditional products that are out there that leverage AI in really cool and compelling ways.
• 56:48
  SPEAKER_00: Excellent.
• 56:49
  SPEAKER_00: Well, yeah, looking forward to uh seeing what what comes down the pike.
• 56:53
  SPEAKER_00: Um if you want to see uh or learn a little bit more about this, um Ari will be uh once again um on Smartstage uh in uh at CD Expo talking about this stuff.
• 57:03
  SPEAKER_00: Um so yeah, thanks, thanks very much, Ari.
• 57:06
  SPEAKER_00: I appreciate your time.
• 57:06
  SPEAKER_01: Let me let me throw one more thing out.
• 57:08
  SPEAKER_01: If anybody um this is not just for Sonance dealers, anybody, if you have questions, you want help, we have two things that you can do.
• 57:16
  SPEAKER_01: We have ai at sonance.com, please send an email there.
• 57:21
  SPEAKER_01: I am part of that group that reads those emails along with a team of people that will uh definitely respond to you and um at times set up a Teams call or something like that with me or anyone else, or you can email me directly.
• 57:35
  SPEAKER_01: It's a R I at Sonance.com.
• 57:38
  SPEAKER_01: So just add an R in the middle of AI, you will get me directly.
• 57:40
  SPEAKER_01: Um, but I'm on both of those.
• 57:42
  SPEAKER_01: And um please, if you are interested and want to learn more or want to get a video of some some presentation that we've done at any one of these uh groups, if it's something we can share, we will.
• 57:53
  SPEAKER_01: Um, the thing with AI, you know, unlike I I spoke I still speak about culture quite a bit.
• 57:59
  SPEAKER_01: That presentation is by and large the same presentation every time.
• 58:02
  SPEAKER_01: And with AI, it has to be different every time because uh the things are moving so fast.
• 58:07
  SPEAKER_01: So um, but nonetheless, I've got some recent things that we've done and recorded, and I'm happy to share.
• 58:12
  SPEAKER_01: So if anybody wants uh help or or just wants to chat about it and um pick our brains, please reach out.
• 58:19
  SPEAKER_01: We're happy to help.
• 58:20
  SPEAKER_01: It's just part of our kind of ethos of authentic partnership.
• 58:23
  SPEAKER_01: We want our customers and the industry to feel like we're a part of their team.
• 58:27
  SPEAKER_01: And uh anything we can do, and anything I can do, just put us in.
• 58:31
  SPEAKER_01: All right.
• 58:31
  SPEAKER_00: Excellent.
• 58:32
  SPEAKER_00: Very good.
• 58:33
  SPEAKER_00: AI at SonyS.com.
• 58:34
  SPEAKER_00: You heard it.
• 58:35
  SPEAKER_00: All right, but thanks.
• 58:35
  SPEAKER_00: Thanks, man.
• 58:36
  SPEAKER_00: Appreciate your time.
• 58:37
  SPEAKER_01: Bye.