Why do 81% of Mature AI Programs Get Rolled Back? with Daniel Morris, CPO at Sinch

(Intro Narrator at 00:00:00) Today, we're talking to Daniel Morris, CPO at Sinch, about why 81% of mature AI programs get rolled back even though 89% of companies are confident in them. You're listening to Joel Beasley, Modern CTO.

(Joel Beasley at 00:00:21) I did get to learn quite a bit about Sinch over the past week or two, and I was curious what your role is at Sinch.

(Daniel Morris at 00:00:27) Yeah, so I'm the chief product officer at Sinch, which means, you know, basically traditional CPO role. I run the product team here at Sinch across our portfolio of communication services. We operate a pretty vast portfolio globally. Simplest way to think about it is we do messaging, voice, and email. So if you're a business that needs to integrate those communication channels into your application, you come to Sinch, and we will send marketing messages, transactional messages.

(Daniel Morris at 00:01:05) Think like your ride receipts after you get off a rideshare or if you're getting an OTP notification, those one-time pin codes that come in. A lot of that traffic goes over our networks for security use cases and things like customer support and customer service as well across those channels—voice calls, messages, emails, and everything in between.

(Joel Beasley at 00:01:36) Yeah, it was interesting because I talked to Jacques, and then I would say yesterday I got an RCS from Verizon explaining to me RCSs. It had these cards in it, and it just had this notification explaining to me how these new rich text messages work. And I said, wow, we are right on the cusp of it. It is being widely adopted, and the carrier, like the users, are actually explaining it to people like, hey, here's why you're getting these different types of text messages.

(Daniel Morris at 00:02:05) Yeah, there's a bit of a transition going on, and we can get into that as we go through some of this shift from traditional one-way SMS-based messages, which are not rich. I mean, you've had SMS plus MMS where you can send multimedia messages, but they were still sort of one-way communications that were happening. Maybe you put a link in your SMS message, but now you can send rich content. So RCS unlocks that. Now you can have actually a two-way communication. You can interact, and we see the same thing happening across other channels. The other big one that's sort of rising globally along with RCS would be WhatsApp via Meta. And it's very regional, like where you see these rolling out. But certainly we are deep within the shift from SMS to rich communications, which is driving a whole lot of new interesting use cases for customers and ways for brands to connect with and interact with their customers.

(Joel Beasley at 00:03:12) Are a lot of these brands using AI agents as well?

(Daniel Morris at 00:03:16) Yeah, certainly. There's quite a bit of this. So we've seen—I mean, you've seen this for a while. I always like to talk about things in waves, so I don't know if this is like a third wave or fourth wave, but AI is not necessarily new. You had a wave of AI back in 2015, 2016, and you had the likes of tools like API.AI, which became Dialogflow, and products like Chatlayer, which is a product that Sinch owns that allows you to build these deterministic chatbots. Right? And people have been using chatbots for years. I mean, what's changed now is with LLMs, right? You can do not just these highly deterministic workflows with natural language processing, but you can do full-on nondeterminism and pointing at your company data sources. They look, feel, and act more intelligent from that perspective. And we're seeing the evolution of companies that were forward-thinking on putting chatbot on their website or in their support portal or where customers are interacting—starting to adopt more of the LLM-based ones for that mix. You know, this mix of sort of determinism and nondeterminism. And now they're doing that, and they're starting to communicate over those channels, not just messaging, but also using them over voice as well.

(Joel Beasley at 00:04:40) Yeah, and I've read your research report, and it found that, what was it? I think the number was 81% of these companies have rolled back their agents. Can you tell me about that?

(Daniel Morris at 00:04:53) Yeah, I think when I looked at that, what stood out in the data is that it's—you're doing AI, you're trying to put these into production. What's actually happening? When you look at the data in customer communications, I think most enterprises, they've crossed that line, right? The real challenge, what we find, it starts after launch. And, you know, if you look at—you're probably seeing this now with AI. Like, it's easy to demo AI. I was watching somebody just yesterday with the new Fable model from Claude was showing this side by side of Lovable, and he's like, he cloned Lovable basically, and he's demoing this. But I think not actually clear that this person, you know, doesn't know enterprise-scale engineering and that, you know, yeah, it looks and feels like a copy and clone of this app. AI is demoing this, but getting that into production at scale is the hard part. And I think that's what we see happening in these numbers, right? The rollback numbers are showing that it's really hard to get things into production. Easy to demo, hard to get into production. And so, I think you see that whether they have visibility and the guardrails to respond before the failures that they're seeing reach the customer. We see that in coming out in those numbers.

(Joel Beasley at 00:06:15) And you guys handle that mostly in communication-related things, but you're not handling that type of thing in normal application development.

(Daniel Morris at 00:06:23) Correct. Yeah, we're focused pretty squarely on the communication use cases and built not only plugging into how our customers channels that we provide.

(Joel Beasley at 00:06:41) My favorite part of the research was 89% of leaders are confident in their AI readiness, and 81% of companies have had to roll back the AI. I was like, those two—I'm not the smartest guy in the world, Daniel, but those two things didn't seem to mesh well.

(Daniel Morris at 00:06:52) Right. Right. Yeah.

(Joel Beasley at 00:06:59) I think that's happening.

(Daniel Morris at 00:07:00) Yeah. If you're talking about it, something like AI was, like, being confident in their AI readiness and—yeah, there's this disconnect happening, right? I think it's a—to me, it's like that's a recent example of that, you know, confidence and readiness aren't really the same thing, right? A lot of organizations may feel confident because they've invested, they've deployed, they have this momentum, right? But they may overcorrect on this. Confidence isn't really protection from things that happen. And I think the report really found that among those leaders who described themselves as confident, you know, they each one of them maybe had at least one rollback that they had to do, right? And I think what we see and what we hear from customers is the real question isn't, are you going to have to roll back? It's more, can you see failures happening early? Do you understand what caused them to fail, and can you respond before they turn into a customer-facing problem?

(Joel Beasley at 00:08:05) Right. And that's where you guys really come in because the failures can happen at the internal company level. And when they hit your APIs, you guys can watch for that so that the customer doesn't experience it. Is that correct?

(Daniel Morris at 00:08:18) That's right. That's right.

(Joel Beasley at 00:08:19) That's cool. It's like a nice little safeguard. It's like, you already need the services. You already need the emails and the SMS, but I'm assuming that's one of the things that makes you guys stand out in the marketplace is your ability to capture stuff before it creates chaos.

(Daniel Morris at 00:08:32) Yeah, I think we—well, it's also that, you know, when you look at whether it's messaging or whether it's email or it's—these are not inherently hard things to do. You send email every day, right? You can send a text message every day, right? The tools are there, but when you're doing this at scale, like these are really hard problems to do at scale and—are your systems resilient? Do you have the monitoring in place to understand when something goes off the rails? Those things are much harder to actually solve and where you can see those failures really being a cost on you.

(Joel Beasley at 00:09:18) What are the costs? Like, what do the costs look like?

(Daniel Morris at 00:09:21) Yeah, I think when an AI agent goes down, what we found, I think, coming from this report, right, the cost, it lands in almost, I would say, probably three places at once. You have your support queue, right? And I think that takes the immediate hit, right? Because work that's happening has to fall back to a human. I think in parallel, when you fail and you're sending people to your support queue, your brand takes a hit, right? Because customers experience the failure as your company, not as a technical issue. It's like the, you know, it's like this brand, you failed. And I think on the back side too, your internal teams, your engineering teams take a hit because now they're getting pulled off of whatever roadmap work they're working on. They're now into diagnosis and rebuild and rework. And I think when you look at the maturity of organizations, some of them are—I think they most understand the support one very clearly, right? Because it's the first line of defense. I think the other two are where the bigger cost finally sits, right? That's where it really hits them.

(Joel Beasley at 00:10:28) And do they—how do they—are they watching these cost in a P&L type of way? Like, are they watching them in dollars, or is it just kind of an abstract thing they discuss?

(Daniel Morris at 00:10:40) Yeah. I mean, I think on the support side, certainly. I mean, most brands, they're following their support queues, right? They understand time to resolution. They're looking at ticket times, and it's a very easily understandable thing that you can do on time to resolution. I think it gets harder when you get into the second and third tier of support, and it's hitting your engineering teams. These are some of your most expensive resources. And I think it's also harder to understand, you know, how, you know—brand measuring a hit to your brand, right? What do you—you're looking at different signals across social media channels maybe and other things. That may not materialize as immediate as, say, the feedback queue that you're getting from customers. But I think, you know, you do see things like reputational damage that comes out of this.

(Joel Beasley at 00:11:33) That's unfortunate for those brands. Yeah. So as a CPO, what is the thing that you're focused on the most right now?

(Daniel Morris at 00:11:43) I think this is important. It ties into this a bit. So what I'm focused on right now is aligning our strategy to kind of three major shifts that are happening in the market. And this is—we talked a little bit—AI is one of them, but there's—it's hard. You know, a lot of times as a business, you might have one thing start to happen, but we have a confluence of three things that are hitting right around the same time. And the first is generative AI. So we're talking about this because, you know, you're asking questions about a recent report that we put out, right? But Gen AI, it's reshaping how businesses discover and experience software. That drives towards agentic. You know, we have builders and developers. It's a shift from—we're seeing applications go more headless. We have API-driven experiences. We have an uprising of conversational messaging and voice. And on that, the second shift that's happening, technical shift, is that move from SMS to rich messaging, and that's RCS and WhatsApp. And they're changing how you can interact. We talked about that in the beginning. Rich, branded, two-way conversations that replaced the one-way SMSs. And then the third is this shift around network APIs, which are opening up a new way to do trust. So I talked about OTP, those one-time PIN codes that you get, which customers use us a lot for. Well, there's a shift happening to carrier-based network APIs that allow you to have new options for identity verification and preventing fraud, which means things like one-time passcodes start to evolve into broader digital trust. Think about this as now you can do silent network-based authentication. I'll get into that. And so those three technical shifts are driving four impacts that I think matter. And the first one's agentic, right? Because agentic changes how businesses discover, how they select, and how they consume services, right? I mean, you can think about this. You're probably hearing about this. You're probably having guests every other day come on the show talk about how it's disrupting their business model. Well, we're not immune to that either. We have to adjust our products to do that. And second, the second sort of business impact would be applications are moving headless. You hear Salesforce talking about this. When the cost of software falls and businesses start assembling what they need on top of APIs, that shifts how you need to build. And we have not just lower-level APIs in our services. We also offer application-based capability within messaging and email, right? Think of it as those simple interfaces to send. And then the third is conversational messaging. So how do we actually change that? And then the OTP, that shift to digital trust, right? When—this one's big, right? Because when you no longer have to send this message, think about the user experience you get when you get those OTP messages. Now you can actually do this silently just based on the device you're holding. So we have products like Number Verify uses this silent network-based checks based on the device you're holding. It's better security. It's a better customer experience, and it's lower friction into what you're doing. And this is evolving now, but it's like—I won't—I can go into deeper detail here, but each of those three technical shifts are driving these four major impacts. And that bleeds into every roadmap that we have across every product in our communication portfolio. And so I'm spending a lot of time thinking about that and how we navigate those shifts and bring the right products and services to solve the right problems for our customers as they are also navigating these shifts with us. So it's like—

(Joel Beasley at 00:15:52) Everybody's doing it at the same time. Yeah. Everyone's—and it just keeps coming. And then it's not going to slow down because the technology is getting even better to make cooler stuff, so it's going to speed up.

(Daniel Morris at 00:16:02) Yes.

Joel Beasley at 00:16:03
So to recap, just so I understand, you're spending a lot of your time looking at how you had this existing roadmap, all these market changes started to happen and technology changes, and you're looking at how all those changes are going to affect the roadmap. Because fundamentally, we were not having this conversation five years ago at all about the ability where everything's going to go headless. You know, because we just didn't have the AI to be able to spin up stuff as quickly as we can now. Like right now, it's almost easier for me to open up Cursor and build the interface that I need than it is for me to spend the time to go try to research 10 different products, try 10 different products, talk to 10 different sales teams, get a licensing agreement in place, find something. Sometimes you'll find a tool, but it doesn't work with your financial model. It's like, I can't do what I need to do with this cost structure because I need a way higher volume. And so you have all these reasons why we should go headless, and now we have the technology to do it. So where do we see the market?

Daniel Morris at 00:17:08
Yeah. And I think what's interesting too when you look at, I don't know if the jury's out on it, but when you watch what's happening from a—so, you know, just being in the communication space and we send communication volumes, be that the number of voice calls or the number of messages or the number of emails that are getting sent. You know, in a world where it's not just, this shift is changing how everybody builds their applications, and maybe you're seeing a proliferation of apps that all need—and they're not static spreadsheets, right? These are applications that are wired into the internet. They are communicating, right? And so volumes start to go up, which, you know, is interesting for a communication provider. And so navigating that becomes one—it becomes a good opportunity for us. But we see the—you also see the volume start to shift out of maybe some traditional applications which may have had that, and they may fan out into, you know, customers are just building their own apps, and maybe they're throwaway apps over time. So the dynamic starts to change as well on where the volumes are coming from. And I think it certainly starts to change when you see when agents—I think you start to hear now people are using agents and they're writing agents to do coding loops and, you know, to build software. But now what about when agents are starting to actually do the communications as well on behalf, connecting into the systems, provisioning accounts? You can start to see where this goes, and you need to adjust and change your underlying systems to handle not just the scale of that, but the change in interaction patterns that are happening within your products. You have a lot, but the dynamic shifts a bit. And, you know, I think we're all still learning a bit how that goes as it's moving so quickly.

Joel Beasley at 00:19:03
Share with me about this silent network-based authentication. Are we talking about, like, is it a passkey thing, or is it when the app automatically reads the text message OTP, or is it something entirely different?

Daniel Morris at 00:19:16
Yeah. I think it's more—I probably will fail to get into the deep technical nuance of exactly how it works, but you can think about it—you have a device. That device has an identifier. The mobile network operators are able to, they understand that device, right? It has an ID, and they're able to identify and verify that device, right? That it belongs to a certain person, right? It can confirm device possession, which improves security. And there's multiple APIs that you can start to expose, and you can use those across both cellular, Wi-Fi. And so, yeah, it just lowers that friction. But think about it in terms of a mobile network operator—a network operator. They know the devices that are on their networks, and they can understand and confirm ownership and device possession, just like when you think about a two-factor authentication. You have another factor of auth, like the device. You're using it already. That device is becoming that factor. But this just simplifies it instead of having to send a code across the wire and then enter the code. You're just using the—the device is doing that, and it's happening silently on your behalf behind the scenes.

Joel Beasley at 00:20:45
Oh, it's brilliant. I would love that. You know, I've always found it odd because I am a Verizon customer, and when I open up the Verizon app, it wants a text password. And like, you know the device I'm on. I have Face ID capabilities. Why am I still—I feel like a caveman entering in this password. You should know it's me.

Daniel Morris at 00:21:05
Yeah. You've already got the device. I think we all share in that. And but I think that's where I think the network providers are going. I mean, there's work to do to build up the infrastructure. You think about it, it gets into, there's a standards that have to get put in place, right? Because you don't want—if every network operator is implementing this in their own way, then you don't have interoperability. So your device as it moves between networks. And so there's standards that get put in place. Those standards then have to get implemented and put out. I mean, the same thing is true of when you look at RCS as well. These are communication standards, and, you know, so it becomes a long road, and I think the industry is working towards that, and a lot of people want to see that come to fruition.

Joel Beasley at 00:21:59
We've talked a lot about SMS, and we talked a lot about email. What's going on in voice? When you guys say voice, what does that mean? Is it like the Eleven Labs AI-generated—like, what is voice for you guys?

Daniel Morris at 00:22:12
Yeah. I think, when you talk to Jacques about this, right, if I remember catching up with him, that voice is—you know, he maybe said voice is going through a renaissance, right? I think voice has started to—it hasn't gone—it's going through that renaissance, right? And it's just become a priority again. I think the technology itself is now becoming more usable in real workflows, and we saw that in some of the reports where we had 67% of enterprises rate voice as equally important than that of text messaging, 41% said it was fully autonomous at handling calls. I think that makes sense because voice interactions are more complex, but you have a higher engagement and you have a higher intent, and it's harder to resolve through simple text exchanges, right? And I think that same rule applies everywhere else. Like, the demo's easy, but making it trusted and reliable at scale is hard. But I think what we saw when you think about what's driving voice, if you've seen the quality, there's the quality of the backend, the AI capabilities that are actually driving richer natural conversations. You know, in the past, it might feel clunky. We've seen improvements in the interaction around latency of those communications, right? So it's actually driving more of a natural communication. You can think about where, you know, when you were trying to do this prior, if the underlying speech-to-text and text-to-speech algorithms were slow or clunky or they weren't translating effectively, it's going to lead to this, you know, really confusing convoluted conversation. I think they've improved to a point now where they're actually driving—you know, it's harder to tell. Like, you can—you're probably interacting now with some voice systems that, you know, it might take you—you probably figure it out, but it's taking you a longer time to really notice, am I talking to a human or am I talking to a bot? They're able to—the technology is able to insert, you know, as I'm talking, I'm saying "um" and "uh," and, you know, they're able to do that. They're able to interrupt within, you know, microseconds of understanding context. And I think that's really pushing this renaissance, I think, as Jacques had talked about it in your last call.

Joel Beasley at 00:24:55
Yeah. It's getting wild. I don't know how I feel about it. You know? Part of me likes—like, I'll tell you, half of me likes it. Like, I'm imagining it's going to get better. It'll be an easier experience. I wish I never had to call anybody with voice. The fact that I—I should be able to do my entire life with text, right? But I still had to call the bank the other day with voice for specific reason. I still had to do it. And I had to deal with humans, and I was just like, I can't wait until the AIs are so good that they can just do what I need to do without me waiting on long holds. At the same time, it scares me because what if I was, like, you know, what if I was in some city that had some very specific view of equality that was implemented in a way I didn't agree with? And I call for emergency services, and they prioritize someone of a different categorization than me for equality reasons. And I'm like, now I'm getting a lower quality of service because this AI just—there's these weird things that could happen, right?

Daniel Morris at 00:26:01
Yeah. Yeah. Yeah. I think, and it's—I mean, you've all—like, too, like you're probably still—I think what's hard too in this is that, you know, we see—you know, being on the—we're pretty—we're at the tip of the spear and interacting with not only the technology that we're deploying, but, you know, the companies you mentioned, some of these ones on the frontier, like Eleven Labs and others. And we're experiencing this technology, but we're also still, at the same time, calling up places and getting stuck in an endless IVR loop of buttons that never take you anywhere. And so we see the potential, but also see, you know, similar challenge in how you deploy that. But I think, yeah, it's here. We're seeing customers, you know, start to engage and deploy this. And I think where we've been too when we look at—we operate, and I think this is what's super interesting because there's a lot that happens at the software layer, but there's also a lot that happens at the underlying network layer. And we operate, in the US, we operate one of the largest independent tier one voice networks, which means we have higher control over, you know, quality and latency of those connections. And so you can actually start to do interesting things in those cases of getting to some of these better, more natural live conversations that are happening. And so it's cool to see some of this work at scale and experience some of the benefit that you can see from this technology working at scale.

Joel Beasley at 00:27:45
One of the coolest things I saw in support over the past couple years, there was a company called Quantum Metric. And they were like session replay, like analytics for your tools. And so when you would call, it's like, say you're having a problem with, I don't know, Salesforce since you used them as an example earlier. When you called Salesforce, their person had the report and knew exactly what thing you had tried to do in your system before you called in. And then they could say, "Oh, I saw you were trying to do this. Let me help you with that." And I was like, that was brilliant. But this is also amazing. Those two things together. Wouldn't that be cool that if the AI could see, "Oh, hey. I saw you were trying to add a plane ticket to Tampa, Florida into your cart, and it failed a couple times," and then the AI could just help you right there. That'd be awesome too.

Daniel Morris at 00:28:38
No, we—I think the—we have a—so I think you're kind of hitting on one aspect, which is interesting when we look at—so across our—it's sort of like what type of intelligent signals, you know, do you have sort of happening across these different communications. And so when we look at—we're starting to look at some things on when you take conversations—they're both multichannel, and they're also multimodal, right? So they're happening—it might happen over messaging and voice. You might have, and then with that, like, you're having two different channels, but two different modes of communication that you're doing. You might be text versus you might be voice. And so you have to—and you're switching across them. Maybe you first open a ticket, but then you actually call into the support line, and then or maybe the brand sends you an RCS message and you actually reply to that. So now you've connected with a brand over three different channels and, you know, multiple channels and multiple modes of communication. And how are you keeping track of what's happening in that conversation and surfacing up the connections and signals that happen across that? That gets really interesting and where you can start to—if you think about interacting with a brand, where you can start to delight a customer with when you have that understanding and signals. And, you know, and today—so we're working on those across, you know, as we think about specifically communications and channels, but we also have a lot—we think about how do we drive intelligence at, you know, kind of every step, right? And so if you look across each of these, we talked about—you know, think about on the voice side, you have real-time audio that you're capturing in there. And so you're seeing those signals, and you can leverage that, you know, with custom business logic to route that to the right person. You could have, uh, your understanding, you know, ownership of a phone number, things like verifying the number, validating an email address. You know, is this deliverable or not? Is this live or real? We have all of these signals. You can look at, you know, things like what's happening in a message conversation, looking at sentiment or things that are coming out that you can bubble up these individual points of intelligence to drive and cater how you want to drive a conversation that's happening across them. And it gets really powerful when you can connect them together. That's also a really hard problem to solve at scale too, but things that we're starting to look at.

Joel Beasley at 00:31:38
Do you guys do custom work with the brands, or do you just provide the APIs for the brands' engineering teams to work with you?

Daniel Morris at 00:31:46
We do both. So we have our products. Customers can sign up as a self-service customer, put their credit card down, start experiencing the products. They can come in and sign a contract to use one of our products and services. And then we also have—we'll work with you on custom solutions if that be the case. We see that really prevalently, I think, in the conversational space because, you know, the technology is one thing, but when you're really trying to craft an experience for your customer, there's a lot that goes into doing that and thinking about the workflows, thinking about the users.

Daniel Morris at 00:32:28: It's very often, you know, it isn't just, oh, let's create a—let me throw an LLM-based nondeterministic agent at it and, like, hope for the best. It's usually this mix of, you know, some level of deterministic, okay, here's how I want to guide this with some, you know, nondeterminism baked in, and then also crafting, you know, the way the messages look, like, how the—you know, understanding how they're received and looking at the journey that somebody might go through when you do that.

Joel Beasley at 00:33:01: What are you learning right now as a leader?

Daniel Morris at 00:33:05: Well, I mentioned before about these three shifts. I think I'm learning about how to drive focus in a sea of what seems to be a sea of confusion and unfocus, because you've got the AI disrupting everything and these shifts that are happening in the communication space. So I think, and with a broad portfolio that we have, really learning how to identify where, you know, where we can win, where we should place bets, where we should stop things, and shift organizations and teams to focus on the right things where we're going to solve, you know, problems the fastest for our customers. I think there's lots of—you know, you might have, you've got to drive efficiency in some area, growth in another area, and you have to make all these decisions that affect teams, roadmaps, revenue, GP, you know, your comfort. And so I'm learning a lot about how to navigate, you know, this, in the best way.

Joel Beasley at 00:34:12: I'm using AI a lot. I have all different types of AI. Grok, Gemini, Claude. I'm using all of them to help me think through things. Are you using AI to help you think through things?

Daniel Morris at 00:34:23: 100%. I think one of the—I don't know when I experienced this, but it was—and then, you know, somebody wrote about how when you go from the shift of, like, I ask questions to AI versus I shift to having it ask me questions. This was a while ago, but, like, that was a big unlock of where it wasn't just me pushing, putting something in, but it was helping to navigate me through a thought process and hone how I think. And I found that that sort of, like, dual dynamic really was a big unlock. I think I've used it as well. Um, you know, I came—we didn't talk about history, but I, in a past life, I started as a software developer early in my career and, uh, so, and then went to the dark side of product.

Daniel Morris at 00:35:14: But I, you know, so I rediscovered, I think, recently as well just the joy of coding. I've been away from coding so long. You know, it's kind of like you don't learn the fundamentals of what it means to program, but, you know, nobody would let me near a production system, you know, writing code these days. But the tooling has made me—been able to experience that, and it brought back this sort of joy of creation, you know, that had gotten lost, which I think, and I found myself burning, you know, burning endless hours and, you know, like, where my wife's coming out at, like, one in the morning, and she's like, are you—what are you still doing up? And I was, you know, working through a problem set and building the tool to solve a problem.

Daniel Morris at 00:36:07: I found it to be, uh, yeah, both joyous and also distracting at the same time.

Joel Beasley at 00:36:13: It is beautiful. I mean, for me, my career has basically been summarized as building the MVP, getting funding, and then building the engineering team. So it was only hands-on code to basically all three products I made. I wrote the code for the first version of the product by myself.

Daniel Morris at 00:36:32: Oh, interesting.

Joel Beasley at 00:36:33: And it is very joyful to be able to build software without having the headache of banging your head against the keyboard because your current, you know, configuration on your computer isn't matching, and the server's not booting up locally. Yeah. Yeah. You know?

Daniel Morris at 00:36:50: Yeah. Yeah. So I can—I can remember, yeah? It's—I can remember endless—I think back in those days, I was working on, you know, some IBM code that was probably millions of lines of code and feeling, like, boxed in at every corner when I was trying to make a change. I sat—I was doing crypto security programming, trying to code different, you know, algorithms and, you know, getting locked in, and to have this powerful tool behind you that can help push through.

Daniel Morris at 00:37:20: It's, uh, yeah, it's been great. Yeah. The—I was just thinking too while we're talking. The other thing that I've been focused on just within—so when you look at Sinch, so Sinch, Sinch grew over the last—one, it's grown a lot organically, but a lot by acquisitions.

Daniel Morris at 00:37:45: I actually came into the company via an acquisition. I was at a company called Mailgun, which became Pathwire and was acquired into Sinch. And, you know, when you do that, it's great. You catapult the company, but you bring in a lot of different cultures and ways of working, you know, and this is globally. And so we've been spending a lot of time lately since I've been in the helm as CPO on how we can sort of transform how our product and engineering teams work.

Daniel Morris at 00:38:20: And there's an AI component to this as well, but it's like, how do we get from, you know, where we've had a lot of fragmented different ways of working into tighter, like, coherent operating model that we can have. And so we've got a—you know, we've started crafting a common playbook, which creates this foundation for how we discover, how we build, how we operate, how we improve our products. And it's super—I bring it up because it's just, it's something that also when we talk about what AI has done, this also excites me because we're starting to see, we're starting to see some of these teams that might be, you know, separated by geography. And as we bring parts of our portfolio together to, you know, bring different messaging products together or bring messaging and voice together, it allows us—I'm starting to see all these little bright spots of how we're solving problems better, we're improving quality, we're learning faster. And I think, you know, we really want to get—you know, how do we get, you know, all of our teams to these sort of fast speedboat type models that are moving through.

Daniel Morris at 00:39:30: And I think AI adds another layer to that because it starts to unlock some of this. And we're seeing—we're seeing all of these, as many companies are, green shoots across the company of, you know, solo, you know, product managers that used to be developers, like, bringing an entire product from inception to, you know, alpha and getting it into beta. And watching this sort of manifest is really, it's really fun to watch. It's an exciting time to be in software as much as it's also an aggravating time to be in software because of all the change, but, um, it's, uh, yeah, it's cool to see.

Joel Beasley at 00:40:08: It's like we're sitting here optimizing and making everything more efficient, and then we have to take a break and be like, wait, how will we get like—how interesting. Because we're a product in someone's portfolio too as far as, you know, the different vendors that they use. And so you're doing two things at once, and it's pretty chaotic. So you have to make lots of decisions. You know? What's one of the product decisions that has been the hardest decision that you've had to make this past year?

Daniel Morris at 00:40:34: It's back to this point of, like, how do—what we have is, like, we're, you know, as a company, you're trying to grow. Right? And, you know, as a public company, you have to generate—you generate shareholder value, but you're also trying to drive, you know, outcomes and solve problems for your customers. And when you have a big portfolio and you have to rationalize that, you also have to get more efficient. And so it's—the hardest things are making decisions to—you know, where you have—think about you have a customer that, you know, and they have a problem and you need—you can solve that problem. Right?

Daniel Morris at 00:41:11: You could go spend time. You could spend resources to do it, but you also have, you know, an inflection happening in the industry, and you've got—you can see out, you know, six months to a year from now of where you need to be in order to capture potentially a much bigger problem. And it's, you know, having the wherewithal and, like, the fortitude to make those trade-offs of, like, okay, we need to slow down here even though that's the most immediate thing. We could go do that, and it's going to, you know, provide that initial bump and we're going to move forward, but we also have this other thing. So it's like making those trade-offs. I think those are some of the harder decisions you have to make. And I think included with that, as you make some of those, it means, you know, you're making organizational changes. Right?

Daniel Morris at 00:41:58: And these changes affect, you know, teams and people, and those are never easy decisions to make, right, when you're doing that.

Joel Beasley at 00:42:07: Yeah. Little bit of a non-answer. No. I'm kidding. I was hoping for, like, we had to let Kyle go. We loved Kyle. That was the hardest thing we've done in the past year.

Daniel Morris at 00:42:17: I thought about that, but then the listeners, they're not going to know who—

Joel Beasley at 00:42:20: They're going to love—

Daniel Morris at 00:42:20: They're going to—

Joel Beasley at 00:42:21: They're going to be like, we love Kyle. Why did I—no. I'm kidding. There's no—Kyle is the name I use when I make a fake male name. Okay. Well, no. It's either Kyle or Greg. I don't know why, but it is. Well,

Daniel Morris at 00:42:35: It's common enough that you're going to, in a large enough company, you're going to hit several Kyles.

Joel Beasley at 00:42:39: I will. Kyle will be sitting there listening. He'll be like, he knows me. I feel seen. We'll start to wrap up on this question. I'm asking you for a piece of leadership advice, but I've got constraints. Okay? Here's the constraint. It has to be something that you learned, you heard, you experienced, you saw, whatever it may be, but you put it into practice a long time ago. And you still think about it to this day.

Joel Beasley at 00:43:05: You're like, that's one of the best things I've ever changed about this. What is that thing?

Daniel Morris at 00:43:12: Well, I don't know if this categorizes as leadership advice, but I think a lot of times, you know, and this will come from personal experience too because I've experienced it and I've put it into practice of, you know, when you build teams and you have people on your teams and they're, you know, they're growing up in their career and they're ambitious and they want—you know, you want to do a job. You want to keep your best people. Right? And you want to grow your best people. And but sometimes, by circumstance, the company is in a certain situation.

Daniel Morris at 00:43:48: Maybe you don't have the role that that person needs or the role that that person wants to grow into, but you want to retain them. And I saw this early on too when I—I remember I was, I can't remember if I was, like, a product manager trying to be a director or whatever, but I was going after a position, and it just wasn't available. And my boss at the time had, you know, he'd sort of said to you, like, look, I don't have this for you. I think you should go somewhere—he's like, I think you should go over here. Take this other job. Go away. And, you know, it was great. And I actually ended up eventually doing that. Right?

Daniel Morris at 00:44:26: And, you know, but we came back together. But I think it's, uh, something that stuck with me as a leader to, you know, to be able to know that, like, you're not—it's not about yourself, it's not about, you know, protecting your team, it's about growing the people under you and giving them the right opportunity. And sometimes that opportunity may be with you and sometimes that may be elsewhere. And, you know, I've put that into practice, you know, a few times in my career, and it's always sort of stuck with me as I sort of felt it myself where I did—at the time, I didn't really understand, like, why would you be, you know, sending me elsewhere? Um, so I thought, uh, yeah, that one's stuck with me.

Joel Beasley at 00:45:08: Well, so if Kyle gets sent elsewhere, we know why.

Daniel Morris at 00:45:11: Oh, there we go.

Joel Beasley at 00:45:12: Yeah. There's Kyle. Oh, Daniel, this has been amazing, man. We made a podcast. How do you feel?

Daniel Morris at 00:45:18: I feel great.

Joel Beasley at 00:45:20: Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you would like to hear discussed on the podcast, either add me on LinkedIn or send me an email, [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.