Miles Ward: You’re really going to see C-suite, the executive leadership at our customers, want to understand the outcome they are going to receive as a result of the investment they make in these technologies and their viewpoint is going to shrink not a year for returns. They’re like every other business. They’re going to look at this quarterly. I think that’s a conversion in the conversation from what’s possible, which is, I think, a [00:00:30] very exciting and a very interesting time, but they’re really going to want to know what’s profitable.
Everybody, welcome to Cloud and Clear. I am Miles Ward, Chief Technology Officer at SADA, an Insight company, and this is a very special episode for all of us. For years we have had the privilege of sharing our time with you, bringing our incredible customers, bringing our partners together to show you what [00:01:00] is possible in cloud. This is going to be the final episode of that series. We have worked for years together to bring all of that to you and we’re just really excited about the next generation of that together with Insight.
It’s really kind of a nostalgic moment for us. I want to really personally thank Tony Safoian. He’s the one who started this whole thing and this podcast. A big, big shout-out to the whole production [00:01:30] crew that’s worked on this for years with us. George was amazing. Stephanie’s done incredible work. All of the investment in time and energy, Alex has been marvelous, that’s really what’s made this special and we’re really excited about that.
I know, Simon, you also had a specific person you wanted to shout out in this thing.
Simon Margolis: Yeah, I was doing a little bit of my homework on where we’ve been with the six-year hundreds of episodes that we’ve got on Cloud and Clear, and I realized that episode number one was Tony [00:02:00] interviewing none other than our next CEO, Dana Berg, so shout out to Dana. He had a lot to do with all the success here. Hopefully we’re making him proud somewhere up there as well for all the work that we’re doing on this, and yeah, I just thought that was cool.
I can’t believe it. I was actually episode number nine way back when, and then I know, Miles, you and I have had an opportunity to host a bunch of these over the years, so yeah, it is nostalgic, but it’s not the end. We’re just transitioning over and we’re going to keep doing all sorts of fun stuff with Insight ON.
Miles Ward: That’s right. The Insight ON crew is spectacular. We’re going to [00:02:30] be a big part of that in the coming years and they have really all the resources to make an even better impression.
So for the last time on this show, I’m super excited to have Simon and I take us out and walk us through not the past of this program, but the future for all of our customers. What’s coming in the next 12, 18 months is probably more change than most of the businesses that we’ve worked with have experienced in the last decade, [00:03:00] and we got to do what we can to help everybody be prepared for that. It’s really an incredible situation and sort of unprecedented, right? I think the running joke is this year we figured out how to make rocks think, and turns out that’s kind of a big deal.
And we have a very unique position in that we get to interact with so many different customers, all of you who are working together with our engineering and professional services teams to build [00:03:30] what’s next. That gives us a really unique lens and a broad view of what’s possible. I think because of that, we feel a responsibility, a debt to ensure that we’re summarizing and bringing back the best of what we’re hearing from customers and what we’re hearing from Googlers and from folks at Anthropic and the other teams that are working really hard to push this ball forward.
I think predictions for 2026 hopefully should help us set the stage for how much you focus [00:04:00] on investing in this new year, where you place your bets, and the kinds of strategies that you use to take advantage of the change as opposed to feeling like it’s going to run you over.
I mean, Simon, you tell me. There’s a lot of this that’s going to be, I think, pretty fascinating. What’s top of mind for you in terms of changes? What’s the biggest thing that’s got you excited?
Simon Margolis: I’d say the biggest area that I’m thinking about that’s making a difference is, one, the advance of these models [00:04:30] is nuts. What we’re seeing happening, what we’re seeing coming out from Google and other players in the market, it’s crazy. And what’s crazy is it’s not, despite the fact that these models are more and more capable, they’re faster, they’re more performant, they logic and reason better, they’re almost becoming commodities, and I think that’s what I’m excited to see more of in 2026, or I guess that’s my prediction.
That’s what I’m expecting to see more of in 2026 is that we’re sort of going to abstract [00:05:00] away the model layer in the way that we’re almost looking at these the way of the processors, which is that most of my customers, when they’re talking about “I want to build a new app, I’m going to put something out in there,” they’re not saying, “can you tell me about the instruction set on the chips that we’re using?” They don’t care at all. They’re probably not even thinking virtual machine. They’re thinking far more abstracted above that. Server list, maybe they’re building a Kubernetes.
That’s almost what’s happening now or I think is going to happen in the model space because where we find the customers [00:05:30] get the most value is not from saying, “Wow, the jump from Gemini 2.0 Pro to 2.5 Pro really made a difference in my life.” It does. It’s a big jump, but the bigger impact is, “Wow, we moved from building all of this manually and laying chain and we had to worry about how we’re building and how are my agents talking to each other” to, “whoa, I’m now building an ADK. I’m using ATA to have my agents talk to each other. I’m using MCP servers. When I want to deploy that, I can put an agent engine that’s a managed service for…”
That’s where [00:06:00] I’m seeing more and more customers get value out of these agentic systems, and oh, yeah, I guess sure, we should use the latest greatest model also, but you know what? Actually, let’s have a conversation about how we optimize and get something that’s maybe more distilled and more performant and maybe at a lower price point. The value doesn’t go away at that point. So it’s a really interesting space right now when it comes to these models.
Miles Ward: I couldn’t agree more with that viewpoint. I’ve worked with a bunch of customers and these models, as they work, there’s really material [00:06:30] financial differences between them. I don’t know if you know, the Gemini Pro 2.5 model, the top-line model right now, its output tokens in comparison to Gemini Flash, they’re 25 times more expensive. So some companies, maybe they’re pooh-poohing that it’s time for optimization yet and you lay out that level of difference and they go, “Look, okay, maybe this does make sense to pay a little more attention to.”
I think that ties back to maybe my number one prediction, [00:07:00] which is really that there is certainly a time for experimentation. We’re very early in this technology generally. This is a general purpose technology. It’s going to take a decade for this to really come to fruition in its full maturity, so I think ’25 and ’24 are certainly years where most of this should sit in the R&D bucket of your P&L. In 2026, that just can’t be the case. I don’t think boards are going to tolerate it. I don’t think shareholders will.
My expectation is that you’re really going to see the [00:07:30] C-suite, the executive leadership at our customers, want to understand the outcome they are going to receive as a result of the investment they make in these technologies and their viewpoint is going to shrink not a year for returns. They’re like every other business. They’re going to look at this quarterly. I think that’s a conversion in the conversation from what’s possible, which is, I think, a very exciting and a very interesting time, but they’re really going to want to know what’s profitable, and the more [00:08:00] that we’re really dug in together with our customers around financially modeling the outcomes, frankly, using these tools to make that analysis more easy and more straightforward for everybody, that’s what’s going to set up folks to transform much more quickly.
Simon Margolis: I think that makes a lot of sense and we’re seeing that already. We’re starting to see the savvier clients start to think that way.
I just had a conversation with a larger customer of ours literally this week, and they were talking about the fact that what they were doing was super valuable to their business. They were taking tons and tons of data that [00:08:30] their employees needed to be able to review in moments when they’re really interacting with their customers and they were using a very sophisticated, complicated, very expensive model to do that and did a great job. Then they found they could use a much less expensive model to do the same thing because it’s tech summarization. They’re not curing cancer in this use case. They’re doing something that’s far more basic, and oh my God, the cost savings that came with that were dramatic.
So what does that mean for then businesses, Miles? I think you probably have some more thoughts and predictions [00:09:00] around what that means for how businesses are actually going to look at thinking about software and as they move forward.
Miles Ward: I think there’s this kind of characterization that’s starting to make more sense. I’m seeing it in a lot of our most sophisticated customers. One way to approach this is to decompose all of the roles in our businesses. Think of not just the… You think of job description for a person, but inside those [00:09:30] job descriptions are the hundreds of tasks that each of us take on, the functions that that role performs. I think there are going to be individual functions for almost every role that get a lot easier to being completely automatic over the course of 2026.
I know of very, very few executives today that are inviting a full-time administrative assistant to manually take notes in their meetings. If you are still doing this, please stop. All of the video chat [00:10:00] tools, including the one we’re talking to you on today, all support summarization automatically as a result of AI. That’s a great example. That doesn’t mean that administrative assistants aren’t useful. They are going to have a whole new range of higher-value tasks to deliver for their leadership, same as every other function in every department for every business everywhere.
That shift where we’re changing the sort of functional behavior in each of our roles is just a huge change [00:10:30] and helping businesses more naturally accommodate change and move through that in a way that is confident and they can adjust their measurements and the rest is critical. I think a way to summarize a bunch of that change is to think of it as different than the way that we’ve described very often the software world, this blurry line between is AI a part of your personnel? I have a great customer [00:11:00] of ours that calls the head of their AI engineering the director for the department of inhuman resources. It’s ridiculously great framing, but that puts it squarely over here in augmenting your staff.
The other side of it is thinking of it as augmenting your software, augmenting the way that your tools work and making it so that they’re more straightforward, more easy. I have seven different ways at SADA to look up information about our customers from our [00:11:30] support view, what happens in our CRM, what’s their billing like in our ERP? What goes on? Where did they come from in our marketing systems? What’s the last time that they came to an event? I could go into each of those individual pieces of software and ask those questions, or I just go to SkyLens, our internal tool chat, and it has all those answers right away.
I think that compression of software into the behaviors, and again, the functions that we want is a huge [00:12:00] step forward in ease of use for many of our customers that are experiencing a vast diaspora of tools, just stuff in all directions, and that ends up feeling like a real way for companies to crystallize the transformation they’re trying to take on is getting closer to the actual work, closer to the outcomes that they want.
Simon Margolis: A hundred percent, and I think that that leads nicely into one of my other predictions, which we’re already starting to see early signs of, which is the democratization [00:12:30] of the capabilities around these technologies because in the old way of thinking, maybe depending on how far, old, you want to go back, you would have line-of-business people that would experience something as basic as an inefficiency in their process and it would be a whole thing. You have to go and do I talk to IT and can I get resources to help me fix this thing and we have to build some software and write some code, optimize this process, which, at that point, you really have to think about is this juice worth the squeeze? I mean, [00:13:00] it can be a lot of effort. It can be a lot of toil on systems and processes just to get that outcome you want.
Fast-forward to today, it feels like we’re getting very close, if we’re not already there, to the point where that line-of-business user does not need to be an experienced agentic developer to go and build some type of optimization, maybe using no-code at all, maybe using plain English to build something that’s going to be helpful to them.
That gets me excited because it reminds me of when I entered the public cloud space. I used to work in the data center [00:13:30] and I was very lucky to have a job that had data center space and sent me into it, and it got me a lot of experience. Had I not had that, I might not have had as much success or experience or be able to be as helpful to the organizations that I worked with. Public cloud was a huge mind-blower for me because, oh my god, with a credit card and $15, I can spin up a couple VMs and start learning how to do things with them and provide actual value and really level myself up.
I feel like what’s happening [00:14:00] now or what I’m expecting to happen next year is that is exploding even further to say that forget about needing to use a VM or writing some software and being able to put it into a public cloud. Now anybody that knows English, or frankly, that’s not even true, anybody that knows their natural language is able to go and build some type of truly agentic system using… I’m thinking of things like Gemini Enterprise, for example, where an end user can go in and say, “Okay, well, my IT team has already given me access to [00:14:30] my SharePoint data, my Gmail data, my Google Calendar data, my NetSuite data,” whatever it might be, and so now I can simply say, “okay, agent, here’s what I want you to go do and just explain it like that,” just like I would to a peer and now I can have a system that’s going to go and build things.
And I think that’s a really basic example. I think the more exciting example is what we’re starting to see on our teams and with our customers, Miles, which is that we used to spend a lot of time arguing over things like PRDs. So we want to build something? Great, let’s type it up, let’s talk about it, six [00:15:00] pages, great idea because it helps make that whole process more efficient. What’s even more efficient than that? Let’s just build it. Let’s just build the prototype.
And so I think we’re seeing more of that now where it’s like, well, forget about needing to go spin up a whole team of people that can digest my PRD and turn it into something and start building prototypes, but before we even do that, we got to argue about all the details. Let’s just build it because now we can prompt something that’s just going to build us that prototype and go. So that access, I think, is pretty exciting.
Miles Ward: I couldn’t agree [00:15:30] more. I think there’s an abstraction that’s a part of that as every one of our different customers thinks through in each of these different roles and functions how they move up a step of the hierarchy in terms summarizing and abstracting these things. I think another framing that I’ve heard for that, that I think will encapsulate a real prediction of mine for the next bit, there’s this great presentation from the folks at Y Combinator where [00:16:00] they say, “Look, there’s two ways you could go about this AI innovation.” One would be to say, say you’re trying to help a legal team. You could try to build an AI to augment and support and you would sell a product like that, a service that helps legal teams to law firms. The alternative would be to try to build a law firm that is an AI system and, yep, it will need some people to be able [00:16:30] to work and operate, but if you start from that mindset, that probably makes more sense and Y Combinator is clear, “Those are the ones we’re going to invest in, just so we’re very direct.”
I think every company out there, if they’re positioning themselves the right way, they have to think of themselves as a brand-new startup that will be building themselves around their opportunity and their problem space over again, and that doesn’t mean tearing [00:17:00] out all the technology that you have. It means wiring it together in a way that’s more serviceful.
So if you think about the functions, the actual outcomes we’re trying to deliver to customers, whether that’s rich and rewarding, super efficient customer service experiences or trying to get through the workflows for purchasing or returns or information-gathering, all that stuff, there are a bunch of software as a service applications that help with that, but what we really want is [00:17:30] the whole end-to-end service to be built out of software. So that service as a software, how do I rotate into agentic systems that are doing a larger and larger majority of the interactions that we currently think of as the work so that all of the people that are working these problems move up to the next layer of value creation?
I think that’s a really useful framing. I expect to see it over and over again across verticals in this coming [00:18:00] year. That kind of building the machinery to allow you to run service as if it’s software, I think that’s a huge change for ’26.
Simon Margolis: I’ll double down on that because I agree completely and you make a good point, which is that the investment dollars care about things like that. So Y Combinator is a great signal because they really do influence the market, and if they care about these things, everybody else should also. But I’m thinking about it from a totally different perspective also, which is that as [00:18:30] a consumer, my prediction is that we’re going to demand this from the organizations we work with. We won’t accept… Just like if I go to a website today and it takes two seconds to load, I’m just going to move on. It’s totally unacceptable.
Miles Ward: It’s not even just that you dislike it, it’s that you’re convinced that it’s broken.
Simon Margolis: A hundred percent, a hundred percent, which you would think if I said that exact same sentence 20 years ago, you’d think I’m crazy because of course, things take time, right? Today, [00:19:00] we’ve expected, okay, with current web technologies, no, it’s unacceptable and something is off if that happens. I feel the same is going to be true for how these businesses work with AI, and again, I double down on your point. It’s not just saying, “Look, we have some AI feature in our business,” but rather truly the business operates differently because of that introduction of AI.
I think about recently I was on the phone with customer service for a vendor I needed to talk to and they had a voice chatbot that [00:19:30] you’re calling into, but it was, I hate to call it this, but it was the old-school version of AI-based chatbot, so it was very much like, “What you just said doesn’t match my five routes, so I don’t understand what you’re describing.” Today’s agenetic AI would never fall into that trap. And so I, as a customer, found myself not wanting to work with this business because they don’t have the type of interface that I now expect from businesses.
And I think maybe I’m a little weird here because I work in this space so of course I’m [00:20:00] keyed into these things, my prediction for 2026 is that your lay user is also going to expect these things. Just like my dad might not be able to articulate to you that he doesn’t feel good about using a web page that takes two seconds to load, but he knows that that’s not a good experience, I think the same is going to be true for even the non-technical folks. They’re going to expect this level of frictionless engagement with vendors and companies, and if they don’t have that, they’re going to move on. So it’s the investment side of it, yes, and you got [00:20:30] to meet your customers where they are.
Miles Ward: I think that there’s a real kind of a transformation in the KPIs that basically every business function uses. They’re going to have to get more clearly oriented around friction and friction-reduction because look, I’m happy to say I’m about as lazy as they make them, but I think I’m probably in about the midpoint for the consumer world. If anything is even remotely difficult, [00:21:00] it’s going to go away.
Now, here’s the interesting bit. Simon and I are probably weird consumers. I type really fast, so I’m super comfortable interacting with a keyboard, but I got lots of nephews and nieces and stuff that are way better in a voice interface, and I got a whole bunch of old folks that frankly would way rather interact with a person or somebody that feels that way in a conversation. So because the ease of implementing these interfaces is increasing radically, [00:21:30] I think there’s going to be an expectation not just that the average user moves through these workflows seamlessly and easily, but that you have a narrowed interface designed for each of the different cohorts of customers you want to interact with. Some people like the font bigger, some people want to hear it in a different accent. Some folks like dense communications versus big verbose explanations.
I think you’re going to have to… The marketers have started talking about it as kind of an audience of one. I think this [00:22:00] is user interfaces for an audience of one, workflows for an audience of one. That level of personalization, I think, really challenges application and experience development teams to think in a complicated manifold way.
I know I spent a lot of time working together with customer service teams who are already like, “Okay, look, we have the digital channel, then we have the chat channel and the voice channel. Okay, those are the three we’re going to manage.” When you have 200 [00:22:30] because you’ve subdivided now to what each of these different ideal customer profiles want to interact with, that’s just a much more complicated, much more challenging optimization environment.
Good news, we have AI tools that help with that, and we’re going to need them. That’s the core prediction for me is that this is out of a great, great… I loved Sapiens by Yuval where it’s… They call it the [00:23:00] luxury trap is another framing for this friction concept is once anybody has experienced that it’s gotten a little bit easier, you’re never going to rewind to the less-easy version. I know of exactly zero development teams that are like, “We are going to get serious about computing our taxes on abacuses just for fidelity reasons. We really want to make that.” Nope, nope. There is an arrow of time. There is an arrow of technology innovation. [00:23:30] It goes this way and the second something becomes doable, it doesn’t take that many quarters for it to become obvious and then it becomes required and then it becomes contractually obligated effectively. So we got to anticipate that the change rate of that is increasing really substantially.
Simon Margolis: Exactly, and maybe this is a topic for a future episode of Insight ON, but what we aren’t talking about is the underlying technologies that are making this [00:24:00] really possible today because I’m thinking back to, this is a while ago, but when accessibility on the web started becoming a big deal, they were big undertakings. We needed to make sure that we had different size fonts for different users. We could do black and white versus white and black to reduce color contrast and things like that, language translations. These were each big undertakings because you needed to go build a lot to achieve these things.
When I talk about now [00:24:30] the fact that when we want to build a new application, we don’t PRD it, we just build a prototype in a couple hours, the same is true for incremental improvements to existing workflows, applications, and so on and so forth. As a result, when you say, “I want to have 200 different ways to interface with a user,” the old way of thinking is, “oh my God, that is a huge undertaking if we’re going to go build all that.” Today’s world with the technologies that exist in these agentic systems, that’s not crazy. It’s not science fiction. It’s very achievable [00:25:00] to go build very bespoke experiences for different customers without a ton of overhead and ton of toil and, believe it or not, without a ton of tech debt also.
So I’m thinking out loud here, but this might be the next Insight ON that we have to go take on.
Miles Ward: I think that’s exactly the right line. It’s not science fiction, it’s doable. I think that’s exactly the spot to leave our Cloud and Clear show. If you take away anything from the last 100-plus episodes of this, [00:25:30] it’s that what we are all working on. It sure sounds like science fiction. When you go home for the holidays this year and you talk about the harebrained stuff that we’re all off building, I’m sure some of your relatives look at you like you are a little nuts, and like you’re saying, you’re speaking Klingon and the stuff you’re talking about comes from the future, but I promise that future, it’s right around the corner and I’m really excited to have Insight partner with you to make that [00:26:00] fiction into a documentary about your success.
So it’s been a real honor to participate, Simon. I’m super pumped that I got to do this, gather with you, and for all the folks that have contributed to Cloud and Clear over the last years, a big, big thank you from our whole team. Also, a big double shout-out to Dana and Tony for bringing us all together to take this thing on.
And look, the journey continues. Please don’t click here. This is the last one of these things. Go [00:26:30] search for Insight ON on your favorite streaming platform. We will be there. It’s on Apple and Spotify, all the platforms. Subscribe now so you don’t miss out on what’s next. I’m really looking forward to seeing you there. Thank you so much.
Speaker 1: If you are making technology decisions that impact people, budgets, and outcomes, you’re in the right place. Welcome to Insight ON.
Speaker 2: We’ve actually seen full-on attacks, full video and voice replication of a CFO basically do a wire transfer of $60,000.
Speaker 3: [00:27:00] AI is many things. It’s not just generative AI, but everybody wants to make it just generative AI.
Speaker 4: And if you are only kind of looking at how can I save money and make my people more productive, you are missing the sort of big transformative opportunity.
Speaker 5: We will also drive $22 trillion worth of growth, really unleashing productivity and allowing us to grow in a way that we never dreamed possible.
Speaker 1: No fluff, no filler, [00:27:30] just the clarity you need before your next big decision. Insight ON, listen and subscribe wherever you get your podcasts.
