Megabuyte CEOBarometer Podcast: Q3 2026

Hello and welcome to the CEO Barometer podcast brought to you by Megabyte. I'm Ian Spence, founder of Megabyte, and I'm joined by my co-host, Neil Aaron Patter, Megabyte's head of research and consulting. In each episode, we unpack key technology, growth, valuation, and M&A trends shaping the UK technology sector. Assess what they mean for CEOs of software and ICT services companies, and share our perspectives on how to turn these trends into practical actions that maximize long-term shareholder value. If you'd like to find out more about how Megabyte helps the boards of UK technology companies visit megabyte.com. In this episode, we'll explain why we believe what we call the AI Goldilocks period is entering a critical new phase. Why anthropic and open AI have unexpectedly launched multi-billion dollar services businesses, and what that tells us about where the value will be created in the AI economy. We'll also explore the SaaS Pocalypse as it reaches its six-month birthday. If you're the CEO or a board member looking to optimize your AI strategy, this episode is for you. Let's dive in. We're going to kick off the show with a look at this idea of AI Goldilocks period and where we are and this idea that we're going through a period of acceleration. Until now, most companies really seem to be treating AI as something that of a bolt-on to existing products and business processes, they're augmenting what they already have. But this phase of what we call the Goldilocks period is that initial phase feels like it's ending. In fact, it is ending. We feel quite strongly about that. AI is now starting to reshape products, workflows importantly, and even organizational design, as we said at the top of the show, and companies that recognize that strategic shift early and make the necessary changes to their organizational design and their strategy, we feel strongly will drive greater share of the value over those that continue to treat AI as just another technology shift. There are a lot of companies that we know that are doing that. Just to start off and think about this AI Goldilocks period, and maybe what I'll do, Neil, I'll think about it a little bit from a software perspective, and maybe throw you over to you to think about it as always, as always, and think about it from an ICT services perspective. But fundamentally, what do we mean by the AI Goldilocks period? It's this idea, and by the way, forgive me if you're a regular listener or if you come to mega right events, you will have heard this a few times, but just to get everyone on the same page here. So fundamentally, what we've talked about a mega right now for a couple of years, is this idea that we're in a period now roughly 2025 to 2030, but obviously it's broader than that, and it's more nuanced than that. But fundamentally, this era where getting AI strategy right and getting AI enabled into your business and your products and services will drive enhanced shareholder value for the next 10 or 15 years. Why do we think that's the case? Well, fundamentally, we're in a place now where previous to maybe 2025, AI technology was very expensive. It was poorly understood. The risk of failure was high, and candidly, the risk of not doing anything was quite low. We've been in a period now where that is not the case, the technology is becoming increasingly well understood. The risk of really embracing it is dropping because of that, and actually the risk of inaction is rising very rapidly. And the example I would use, all the kind of comparison I would use was the is the SaaS and cloud era 2010, roughly 2010, 2015, where we saw companies, whether they were startups in that era or whether they were existing companies making a change, those companies that got that really right and really lent into that strategy during that period have quite clearly, for the most part, driven a much stronger shareholder growth and competitive positioning and therefore, shareholder returns over that period. So from a software perspective, we can get into it, but from a software perspective, that has been all about the partial agentic, getting agentic products and getting agentic technology into products and services and starting to think about AI automation internally. But again, talking about this idea of the first phase of that, it's been very much adding on, adding layers. And fundamentally, what we see now as an acceleration, I'll get on to that and get on to why it's changing. But what are your perspectives of this early stages from an ICT services perspective? I think it's, as always, more complicated and more nuanced because ICT services, we like to call it one bucket for the reality as it's many different business models, many different areas of the market, many different types of businesses. And so I suppose the underpinning AI element to all of ICT services initially was, as I've talked about with this podcast in various previous episodes, is that that whole operational model, how do you bring AI into your service desk if you're an MSP into your delivery mechanism, if you're a consultancy, do you bring AI or IP more broadly if you're a reseller? I think it's now, we're still in that conversation, don't get me wrong, there's still lots going on there. But actually now, we're talking a lot about the product service strategy for an IT services business. That's kind of quite new, isn't it? There's been a sense that the service, for CITMS anyway, the service desk and all of that automation to some degree is the product. And I think that's still true. Yes. But we'll get on to this later when we're talking about the AI services ecosystem developing. But I think that's a really interesting development. But I think it is, and I think it's so quick, if I'm honest, we weren't even talking about this six months ago, so even two episodes ago, this wouldn't have been part of the conversation, we would have still been talking about operational model and operational excellence, and still we're 5% of the way there in terms of the market. But product and service now means lots of different things. I won't do too much spoilers in terms of topic two, but the reality is that if you're a services company already, we're starting to see different market drivers trying to force you or push you towards having a product and service strategy around AI. What does that mean? For us, we talk about AI enablement services. So there is the idea that you provide the underlying plumbing to help organizations drive an AI strategy. In some cases, it's about actually deploying AI solutions, and there's a whole piece there about how do you help organizations actually deploy AI, whether that's a agentic workflow, or it's a chatbot, or it's actually a full function. There's lots of stuff going on there. It does depend on whether you are a enterprise, public sector, mid market, SMB, there's new answers there. The reality is, by the end of this year, if not sooner, you should have a fully fledged out idea of what you need to be doing. There's been a significant amount of change, and that's been very quick, but it also will massively impact how you think about investment, how you think about the ROI behind that, what you need to be investing in. All of that needs to be built out, I think, this year. I think the word I keep coming back to is acceleration. It's worth spending a minute, just thinking about, okay, we talk about this acceleration, but why? Why is that happening? I think there are probably three or four things to sort of bring out. The first and probably the most important is massive improvements in the underlying tech. Most of our listeners will be the same as us at Megabyte, where the way we're using AI internally is just transformed fundamentally because the technology is transformed. Fairly fundamentally, particularly, let's be honest, anthropic and flawed, but across the board, and that has enabled so many more things, and so many more things out of the box than was possible previously. Things like interconnectivity with systems and a whole load of other things that have just improved massively, I think has been a key, a very key driver. The SaaS apocalypse, as we talk about, and we're going to spend a bit of time talking about that later on in the show more specifically, but it's been a massive wake-up called particularly the software companies that I think were fiddling around with this. If we're really honest, there was quite a lot of slideware going on, AI washing, all this stuff, and everyone is just real. Well, the best CEOs have realized that even if they were doing that, that is just not going to cut it, and they really have to take this very seriously because their valuation fundamentally depends on it. FYI, I've been saying that for years, but it's never good to be too arrogant about these things. But that SaaS apocalypse has been a massive wake-up call. The cost of AI, thirdly, the cost of AI, as we all know, is going up very rapidly, but I still think we've just seen the beginning of that, and I think that is forcing people to think much more carefully about returns on that spend, because when it's diminimous, they're like, oh yeah, you're not yourself out, see what you can do with it. When it starts becoming material to the PNL, the CFO is going, okay, this is great, but actually, what am I getting for it? And that's not always clear, so I think that dynamic is forcing a bit more traditional business rules into it. And actually, and I wouldn't overstate this, and I'll bring you in on this actually, from a services perspective, but I do think we are starting to see customer pull now. There's an education process going on in the customer base of the software and ICT services businesses that we track at megabyte. And there is, I mean, some of it's still pretty generic, I think, I need to AI, rather than being very sophisticated, but that is, I think, making, again, is an accelerant. People's thinking about their products and service strategies, actually, I'm getting customers asking for this now, and I really need to get into it. So, I mean, I don't know what you're, I mean, any of those things, I mean, particularly on the last one, as you've got any thoughts on how that is driving services, kind of adoption. Yeah, I think the latter one that you mentioned there about the customer pull, I didn't expect it to happen this quickly. I'd expect it maybe, you know, initially, if you're maybe targeting a large enterprise or public sector, when you're filling out your, you know, your tender forms or whatever, there'll be a question in there about how you're using it. Obviously, that's happening, so that's not new news. And I thought that's maybe where we would be, for quite a while, I thought SMB probably wouldn't touch this all that much in the grand scheme of things for a while, maybe until 2027, but as people listening to this will probably know, we do a lot of customer everything. We do a lot of customer sampling. And what is now becoming really clear is that even in the SMB space, but and even more so in mid market and to probably the customers are now starting to ask, not just about what are you doing with AI internally. I think the question there might be aimed at how are you reducing costs in your own business? If they don't care what they're doing internally, they just want to know what they can get left for a lower cost. Exactly. You've got to be ready for that question because I think that is happening. But actually, what's been really interesting, particularly over the last couple of months, is that we've talked to customers and they've talked about their service promoted and gone, yeah, they're really good at what they do for us, IT support, security, data, whatever. But we know there's all this stuff going on with AI, both as a generic sort of the frontier labs with their model with their LLMs. But we also know there are AI startups in particular applications, or software, or there's certain. I didn't mention that, so to cut across you out. I didn't mention that. But actually, this idea of increased competitive pressures, wasn't on my list in my notes. But actually, I think that's. Whether it's competitive pressure or just customers are starting to understand that there are these solutions out there. So, I know, let's not get too dramatic about it, but let's say a Salesforce or a HubSpot versus an AI native CRM, for example, they know that these things are going on out there. They know that their competitors, their own competitors, potentially might be using Claude, or they might be using OpenAI, or they might be using Copilot, or whatever, to build applications agents. And they're set there going, I know this stuff is happening out there, and my service provider, who basically might be my IT team, or supposed to be the subject matter expert, is not coming proactively and talking to me about this. If I ask them, they might give me an answer about why I need Copilot or whatever, because I know they've got their Microsoft partner. But no one's coming to me and talking to me about this. And I think this is starting to really affect customer sentiment. It's not like they're going to leave as a customer. In the short term. Yeah. But the reality is just like when they had questions about Claude and public cloud, and they're going, "I'm currently on-prem or I'm currently in a hosting solution, I've heard about Azure and AWS, should I be moving to that?" And their service provider didn't have a very good answer for that. This is the modern equivalent or the today's equivalent of that. So, coming back to your point earlier about this, their history is repeating itself. This is where we're at. So I'm not saying that you must have the perfect answer today. I'm not even saying that you must have a set of products and services today to be able to answer that. But if you can't have those conversations and you aren't clear enough in your own strategy to be able to go proactively to your customer base and go, "Here's what we're seeing out there in the AI world. Here's what we think is relevant for you. We have certain services around that, or at least we are developing a set of services. And this is our roadmap around it that I think is generally starting to affect." And this is where your competition, especially that maybe the newer Gen MSPs or consultancies, the data consultancies, the companies called themselves AI consultancies, this is their go-to-market strategy is to come in and go, "Your existing service provider isn't talking to you about X, Y and Z because they don't know it. I know it. I can do a much better job for you." That might start with them actually taking those service lines or those opportunities off you. But their strategy is to take that starting cream off the top of the milk and then go after the rest of it afterwards. That is the classic wedge strategy. And that's exactly the same in software. Feature kind of quite point solution, almost task-based, quite simple solutions, either from another vendor or internally developed using on top of Claude or whatever you're using, other AI, LLMs are available. That wedge strategy, the U.S. Coast VCs talk about it all the time and it is massively. So the example might be you might have a Salesforce instance and you were previously using Salesforce for your, obviously your system record for CRM, but also all of the workflow tools that they have and potentially their marketing optimization tools, their mile tech stuff. And progressively you're probably still going to use Salesforce as your system of record, but progressively what these, still we're getting to the SaaS apocalypse stuff now, but that system of record still remains, but all of the workflow stuff around it is potentially replaceable. And so that has massive implications for NERR for those companies as they, for the likes of Salesforce where they think, oh, we keep the customer fine, but they're spending half what they were spending before because they've got all these, they're not, then we're not using the mile tech stuff because we're doing it well. So there's a similar bit different, but that wedge strategy was about selling more modules on top of the course. Exactly right. But now if they're losing that, so their NERR becomes NERR, I guess, yeah. Or actually, their NERR is lower than their GRR. Is that possible? That's not possible, is it? They're, their customer attention rate probably isn't too bad, but their revenue retention is, is a shocker. Yeah. Anyway, there's, again, we'll cover that a bit later, but that wedge strategy, I think it's interesting to take away here because that's, that this happens, this stuff, this stuff happens slowly then quickly. Yeah. And it's happening in the SME space. So there'll be people sort of sitting there going, you know, I, I do IT services for barrage fish and chips, they're not really talking about this. They might not be talking about it because they're sort of sat there waiting for you to come and talk to them. Now, they might not take anything out. So I'm not sitting there going barrage fish and chips once an AI forward back office function, but their service provider, whether it's a consultancy with a professional service with managed services, they're looking to you as the, as the source of knowledge on this and you should be able to go and have these conversations with you, you might actually have that knowledge in your technical basis, just not making it through your account management function to an end customer conversation. So it might just be tweaking here and there of how you engage with customers because you might have the technical knowledge. But if I'm honest, I do think a lot of MSPs in the UK actually don't have the technical knowledge either. They're good at IT support. They're good at Microsoft support. But do they actually have the skills required to have these conversations with customers? Do they have the end user knowledge? So the end market knowledge to understand the problems that a law firm will have and as a result, what are the sort of technologies that you can deploy into that? These are all the existential conversations that we're now having with management teams who are tackling mid market SMB enterprise. Because that's actually step one is understanding where they are in their technical knowledge, what they can then do about it from an account management perspective, and then as a result, what their products are actually, product service strategy needs to be. So I mean, I think we've probably covered a lot of this, but I mean, just to sort of really ram the message home about the acceleration in the evidence, what's the kind of evidence? How are we backing this up? Well, I mean, we're backing this up the dozens of conversations we have with CEOs in the UK scale of mid market sector, but there's also endless surveys, ONS, Deloitte, UGARV, I found a whole load of them when I was prepping for the show, just demonstrating that AI adoption is accelerating. So I don't think that's really open for debate. I don't think. I think the interesting one is, is this idea of AI moving up the CEO agenda? And I kind of did different different takes on this, I think. I mean, it clearly is in the sense of going back to the, going back to the drivers, who talks about the beginning, like if you're particularly the SaaS apocalypse stuff, I have definitely seen in the software sector, CEOs that have really, it's been a massive wake up call as I said earlier, and they have lent into this very much, and are very much leading the charge on this, if they weren't before, but you had an interesting perspective on this when we were talking about it the other day. Yeah, so stealing a line for my favorite podcast, disagreeing a group. Your favorite podcast. Sorry, other than this one, obviously. Yeah, so I think you were talking about the fact that, you know, AI as a, let's say, a concept or an adoption concept was moving up the stack to, to the CEO. And I made sort of, not disagree with the opposite view, which I'm saying is actually coming the other way. It's actually a top-down approach, because you certainly, from my experience, there's a lot of CEOs talking about AI and go, we want to be an AI first company or whatever, because that's lovely marketing. But actually, the rest of the organization's going, "Oh, he's going off on one again, and yeah, I'm sure we'll get, this is another phase." But actually, what I'm seeing is that, actually, that adoption is going from the CEO to whatever the SLT is now to users and employees at all levels. And I think the reason for that is because of what you spoke about earlier, the technology has moved on to the point that very easily, people can see some improvement or just the smarts are more available to an employee. But also, I think, because that we're, it's just more widespread, people are able to see, I don't know, someone doing their job in a different company, what they've been able to do with their eyes. So now they're going, "Oh, well, I want to be able to do that." Oh, my CEO's been talking about AI and how it can improve productivity. Now I'm finally starting to see that. So I think there's an element of the market's just grown up. You're also a bit of circularity around it. But it is, I think it's probably, it's both this happening. It's going from bottom up and a top-down approach, but it's interesting to see that, yeah, that top-down is certainly quite apparent to me. And I think, I think, you know, when we try and sort of synthesize how this is manifesting itself in trying to break it down a little bit, you know, I think we've moved from a stage where this is an interesting new thing. AI, it was an interesting new thing to, "I understand that my shareholder value depends on it, depends on me getting this right." And I really mean that. I mean, we use it so we throw around the term "sheld of value." And I use megabyte as a kind of example here, you know. Lots of companies are looking, actually, this is, this is really, I understand, this is existential. In our case, it's like, we're fundamentally content. Well, we're not fundamentally, we're lots of things, but we're events, business, et cetera. But content and proprietary content is the core of our value proposition. Is that going to remain proprietary? Now we're comfortable with that, but we've really had to think about it. And I think that kind of really thinking at a deep level. And then, so this, how does that manifest? And I think there's kind of five things I would highlight. And we sort of touched on these, but to try out, I think it's quite helpful to just really think of them very clearly. So you go from AI features to a deep product strategy. You go from individual productivity to workflow. I think that's a really important one, ultimately, then organizational design. And it's that last bit, I think that really is probably the next step for a lot of people, even if they've made the second step. From permissive to managed adoption, I think it's super interesting. That's a little bit your top down kind of cycle theme. So the companies where the CEO or the leader in the business gave permission effectively for everyone to use it. Everyone's gone crazy now, which is a good thing. But then we need to complete that circle and then make sure that people, I mean, there's a governance aspect to it. There's a cost aspect to it. That's all. And that's, I think, a key focus, increasingly a key focus. Okay. Everyone, I've got this. I've achieved what I wanted to from from the beginning and get people using this stuff. Now we need to work out what's working, what's not working, what it's costing, making sure we're not breaking any rules. We're not leaking data onto internet, etc. From technology ownership to business ownership. Interesting one, this one. So was it the CTO's job? Probably if it was anyone's job, it was maybe the seat in terms of implementation. But it's now I put CEO in my notes, but actually business ownership. This is a business issue. And I think the companies that are getting that right are the ones that are on the right track. Understanding this is yes, of course, it's technology, but fundamentally you can get the technologies right as you want. But if it's not embedded into the business, you're not going to rest value from it. And lastly, from universal excitement to selective conviction, I quite like that one. So again, it's a bit like that managed adoption thing. Everyone's terribly excited about what this can do, but actually working out what you really should focus on because you can't do everything or you can't do everything economically and you can't do everything within what's right and appropriate for governance perspectives is becoming the kind of key features now. So I would say those are the those are the main features of what we're what we're talking about. I think it's, you know, using us as an example on the shame at least, you know, oh, we've gone and created loads of dashboards and stuff, which is really cool, because I can do it without any sort of coding knowledge. And you sort of sit down and you go, yeah, but what can you now do as a result of that? What changes? What changes? Well, how am I going to do business better as a result of this? Yeah. And it's very great. Fantastic. You've done the right thing there, but actually asking that question rather than just going, I can do cool stuff with it. And you're like, okay, fine, cool stuff is only called as useful. So I think, yeah, we've definitely got to that stage. So I think we promise we promise on the show to try and do we use the term actual insight as much we can, but we do really try and focus on that. So before we move on to talking about the services ecosystem and changes there, just try and pull this all together and and give, you know, our listeners a viewers listeners a sense of really what tangible things they might be able to do to sort of take advantage because we're occasionally occasionally accused of being a little bit negative on the pod when we're talking about the kind of outlook because it has been a tricky couple of years. But to be more positive, what can you do? And I think fundamentally, not thinking about what's your AI strategy, but how is AI changing your strategy if that nuances clear? And I think there's a few things I just run through. I think as a sort of key takeaways, we wanted to get across before we move on to the next section. Stop thinking about AI is another product feature. I think we've covered that, but it's so important to think of AI as how it fundamentally changes your product strategy when you're thinking about customer facing. When you're thinking internally, not just about how AI automates tasks, but the real opportunity is redesigning end-to-end workflows and ultimately parts of your operating model around AI. And again, that's quite a shift and it requires, in my view, proactive thinking about how to do that. It doesn't necessarily happen within your existing kind of strategy development, your QBRs and that sort of stuff. You have to think outside of that all-term, outside of the box and actually make it happen. I think this is probably where it's going to interrupt your flow, but I think this is where the top down becomes really important because AI is not going to come as a surprise, but that's going to be where the greatest organizational obstruction will be because if you're asking people to kind of redesign workflows or entire functions, that's then where you get into sort of office politics and corporate politics. And so if you ask someone to just redesign their entire department around AI, that becomes, okay, what he talks about, jobs, roles, tasks. So this is where really clear direction has to come in because if you just sort of set a rule of, you know, we must be using 20% AI in our finance department, people will find ways around that and actually won't get the transformation you are looking for. So some real clarity about what you're doing and why you're doing it. I think it's going to be important because otherwise you're just going to get into this horrible state where people just won't adopt it and they won't do it. Yeah. And I think that moves on nicely to the next point, which is about encouraging, moving from encouraging experimentation, which is still important to actively managing adoption. And I think again, that's a process change that needs to be proactively thought through and pushed through again, leadership. I still think there's room for using your employees as the whole sort of slightly hack me thing about your younger employees, but it's definitely a thing that there's more brain space and just drive within younger employees to do some of this stuff. But you can't rely on that. You need to just continue to encourage that, but you need to operationalize innovation with AI. That makes sense. And I've said this a couple of times, but recognising that AI is no longer a technology strategy. It is driven by technology, but it is a business strategy both internally and for your products and services. And it is a CO and board level issue. I think we've covered that. And last but definitely not least becoming increasingly selective about what you're going to do, pick your battles internally and with your product. The air of adding AI to everything is over really. It's not affordable anymore, even if it ever was. And the winning companies will understand where they're going to drive value. And we'll talk more about this a bit later, but I'm afraid one kind of elephant in the room is some of this stuff is, how am I going to save heads? Where am I going to? A slight different point, actually. But being selective means how am I going to drive tangible, measurable value in the form of increased growth from a product and service perspective versus base and lower cost or lower or higher margin or whatever operational productivity looks, looks good. Your strong productivity looks like for you. And then I suppose to your point about people is then what does that mean for my recruitment retention strategy? So that's probably where it's you do all of that first and then you go right based on that what is my recruitment retention strategy? What skills do I need? What qualities do I need to recruit in or what do I need to retrain my people existing people on? I think that's yeah, but you have to start with that product and the ROI and being selective. Okay, so shit's getting real with the AI Goldilocks period. Hopefully we've conveyed what we think why we think that is and what we think you should do about it. So that's the end of the first section. We're going to have a quick break and we're going to have our own little hydration break. We're going to learn from the from the world cup and we will be back in a minute to talk about frontier labs and you're going to lead on this nail and what that means for services ecosystem. Okay, so I think I wanted to talk about a slightly different topic but all very related to AI I suppose in the grand scheme of things and this is the big news over the last couple of months since we recorded the last podcast, which is around the AI frontier lab. So there's the big guys open AI andthropic and how they both dramatically entered the services arena actually shortly after we recorded the last podcast. So in the space of a few weeks, both anthropic and open AI have stood up basically multi-billion dollar, private actually backed JVs aimed squarely at getting AI and agents into the guts of basically businesses from a services perspective. And then what happened very quickly after that was they weren't alone for long because within weeks Microsoft then launched their frontier services company which they talk about 6000 strong FTEs. I'll talk about FTEs in a bit. And then AWS actually all the announcements got drowned out by the two frontier labs and by Microsoft but they also committed basically a billion dollars to the very same forward deployed model. The thing that is important for me is that this is the clearest admission yet that the frontier labs and now the cloud giants as well see the primary value of AI being in its deployment and that driving this deployment is thinking from part of their long-term growth plan. This is really what they're betting. And obviously they've got their technical strategy around the model and the war they have with everyone else on building the best possible model progressing that as quickly as they can, competing with the Chinese who very recently have announced an unbelievable model at an unbelievable cost base. So there's a big competitive strategy around that. But then for them if they're thinking about you know with open AI trying to list obviously that's no secret anthropic won't be far behind them in that respect. Microsoft is obviously trying to drive adoption of Copilot and AWS is obviously trying to get a horse into this race. For them they've clearly made the decision that the greatest driver of their AI model effectiveness and their long-term growth strategy is making sure these models are being used in actual real world applications in enterprises. I suppose the big question for UK services companies and management teams is this a threat and this is the when we were prepping for this there was a big conversation between us between is this a threat or is it an opportunity for the UK mid market for our sort of core listener base. I think what we let's what we settled on was and we'll talk about sort of maybe our views the reality is this is both some will find that they're relatively undifferentiated and surface level capabilities are massively disrupted because actually they're competing against the makers of the models and therefore deployed their forward deployed engineers and frankly they just can't compete you know whereas others and our view there's a lot in the UK and there's a lot to be celebrated in the UK who will have best practice they'll have the next-gen talent, they'll have market familiarization, they'll have the subject matter expertise, so actually they're able to use best practice. And customer relationships. And it will be actually a boon for them for this market. So I think that's what I wanted to discuss in this part of it. But I think really for me, this idea of deploying AI in the right way into real world applications and corporations of all sizes is going to be where the AI battle is won or lost for services companies. That's really what it comes down to. So I suppose to discuss, from my perspective, especially when I think about the UK mid market, the first question we've got from quite a lot of people is, are these so owed from Anthropic Deploy Co creatively named by OpenAI, but then Frontier company by Microsoft, are these a threat to our core coverage, the UK mid market? I think for me, in a way, yes, because I think what it will highlight and what I talked about in the previous topic is, how does this affect customer expectations? Do customers see this and see what ODE is doing for maybe larger organizations and go, well, why is my service provided not doing this for me? Do they have the skills? Do they not have the skills? So I think it does form a threat to them in that way. I think if you're directly, if you're unlucky enough to be in the kind of market where you're going directly against these Frontier labs and their services, JVs, you might be in a position where your proposition looks quite undifferentiated. It highlights, you don't really know the models, you don't know how to deploy them. You might not really have the customer relationships to do that. So you're on an even playing field with an ODE, which has unbelievable resources behind it to go after you. I think in that situation, it is a bit of a threat, but I think you had a more positive take on all this. Yeah, I mean, I thought for a while that just looking at them, the growing complexity, just on a fairly simple basis, the growing complexity of all the models and all rest of it, it kind of felt a bit like to me when everyone started to talk about AWS 10 years ago when just the complexity of the product, not just from a technical perspective, from a commercial perspective, got increasingly complex. And that was a catalyst I felt for everyone to go, I need a partner. And because in the early days of AWS, no one used a partner for AWS in the early days, when you were using AWS as your mostly software companies, using it for your dev environment, no one was using it for production initially. It was only dev environment. It was all direct. You didn't need a partner. And it kind of feels a bit like that. There are some important differences, which we can get onto. But so I felt for a while that the LLMs, the Frontier labs, will want to and need to use an ecosystem of partners, whether it's traditional channel with a share of revenue, whatever I don't know, but we'll need that partner ecosystem really to propagate the products into the, particularly the midmark NSME. And so to me, this is just a kind of staging pose for that inevitability is probably overstating it, but that likelihood, I just think it's a massive in some ways. I was just thinking about this while you were talking. In some ways, it's the equivalent for the services ecosystem as a SaaS apocalypse, although not as negative as a SaaS apocalypse, but it's that wake up call to say, look, this is the signpost to the future. This is what everyone's going to need. And the LLMs have worked this out and they need to, they need, they need services companies to make this all happen in a business to business context. They've worked that out. So I, I see it's really, yes, of course, there's a more a competitive angle to this undoubtedly. And if you are coming up against these guys, I mean, I suspect in the scenario you just talked about the mid market services company will be a lot cheaper than owed or deported, so there's that's a strategy. But notwithstanding that in the mid market in the SME, I think it points to following, following the hair product strategy and service strategies, you could do a lot worse than doing that and then deploying it and using your AI mode, talking about AI mode in the next section, but your AI mode is a services company with vertical expertise, customer relationships, all these things we talk about to do the same things effectively in your world is a massive opportunity, I think. And to put a few numbers, I think that point there about the services opportunity to tackle the complexity, take the stat with a pinch of salt, because absolutely no idea how anyone would even figure this out. But the, the, the stat that I got was that for every one globally across all customer segments, for every one dollar that's spent on software currently, companies are spending roughly six dollars on services, whether that's implementing it, delivering it, supporting it, the entire ecosystem and the infrastructure around actually making software work. Now, obviously, there'll be a big skew towards the enterprises there where obviously they've got very complex software. But I think the point here is that there is for every dollar of spend on an AI model, which is now what we're talking about, there will be a revenue opportunity around the services to support it, make sure it runs correctly, make sure that it is specific to what you actually need it to come back to your point about ROI. Companies are already getting to a stage where they can't just deploy AI for the sake of it for no, with no ROI, they need to make sure it's targeted that it's correct, it actually drives ROI in their own business and the governance piece and what they need is a service partner that can help them make sure a service partner comes to them and goes, okay, you want to deploy Claude, we would recommend doing it in these three bits of your business because it works best for that and we know that if you do it, you will get a 15% ROI on it or whatever and that meets your minimum requirement. Without that services piece, I think obviously people will be deploying Claude, people will be trying stuff, there will be lots of vibe coding going on but I think that ROI piece is much more difficult with that. And that's development as well. We're all amateur coders now while I'm not, but that goes to take you so far but ultimately if you don't have internal dev function, which most organisations still don't then you do need that support and that full circle on everything, software was all productised in the pre-sass era and now there's more kind of self-developed software coming through. I think there's also a positive, and maybe I'm reaching here, but no, I don't think I'm reaching, I think we can see it. I think there is a massive seeding conversation to be had about this so the two labs themselves have basically committed, committed is not the same as deployed, but committed around five and a half billion dollars, so owed is about a billion and a half and deploy codes about four billion. They'll do that through acquisition, they'll do that through organic development, so owed actually from anthropic came as a result of an acquisition they did in May and then deploy code, they bought a company in the UK called Tomorrow, which tomorrow, and that was the fundamental starting block for them. Microsoft have committed another two and a half billion, AWS has committed a billion. Obviously, big numbers there and you might say they can go, "How the hell do I compete with that?" Then the reality is that they're deploying that with also the idea that they're going to be seeding quite a lot of different things into the market because again adoption is what they're all aiming for, so whether they do it themselves or whether they seed an ecosystem for the future, that sort of roughly nine billion dollars I think has potentially some effects, A, if it's therefore deployed engineers, so this is the Palantir model, which is fundamentally, you have engineers, you deploy them at the company to learn the business process to improve it with Claude or whatever, and so if you're deploying those engineers, at some point they know they're going to leave, they're going to leave, start their own consultancies, they might move around, what you're doing is seeding an entire market of skills that can be used either by the big frontier guys or probably more realistically start spreading out and seeding into the market, so I think there is as much potential for the UK as there is any other market for those FDEs to start moving across companies, so this comes back to your recruitment retention strategy as a services company, what kind of qualities are you looking for, what kind of skills are you looking for, are you going to start, do you have FDEs in your company already that you don't really think of as an FDE, could they be deployed in that way, could you train people in a way that they get the skills that they can become an FDE, or in the future do you recruit FDEs, and that's kind of skill set that you do, but I think also it seeds the thought process if these companies are starting to drive applications, it's like to build applications in large customers or large enterprises in our view, if they start to do things using this model, is that then learned by the mid market and by SMBs as a best, the new best best process of the best way of doing something, does that seed it, I think that's also the case, and then finally I think something more specific I suppose to service providers is the channel, does this start to create a new channel, so even though these frontier models are talking about their own services arms, the reality is that they're going to be partnering, so even Microsoft when they talk about this 6000 FDEs that they want to build, a lot of those, and I've spoken to quite a few Microsoft partners about this, is actually the reality is that they're leaning on their Microsoft partners who have professionals and technical professionals, rebadging them FDEs and basically using them, so that's part of the Microsoft, and that's fine, I think that actually then leans itself to this idea of it's an opportunity, so if you're a Microsoft partner or if you are one of these new sort of Claude and Thropic partners, OpenAI has a partner model as well, there is a lot of opportunity here to get the right skill set, being the right part, being the right rooms, so actually isn't about being left out, or actually these guys eating. your part of the pie, arguably the pie is getting larger. These guys have made a big bet that services is where they want to drive long-term growth because it drives adoption of their model. And I think that's only an opportunity, whether you are dealing with various fish and chips as an SMB or you are doing mid-market, as we define it. Anthropic talks about their mid-market, but I'm pretty sure that's still our definition of enterprise, but the reality is I think there's opportunities all across the stack, which is, which is finally, I think, positive, but you've got to get your product service trashy, right? Because if you're sort of sat there going, yeah, we do AI. We have co-pilot. I think that's just yesterday's trashy. And so what does this all mean for the market then more broadly? How would you summarize that? I'd say there's a couple of things, there's a couple of things, maybe to-do's, I suppose, if I can be so arrogant as to try and give some to-do's. I suppose it is, think about your vertical or your subject matter expertise. I've talked about this multiple times, but this is even more important now. If you've got subject matter expertise in a vertical or a text, it's a matter. It is a matter. It is a matter. It's crucial. If it's just generalist, it's going to get competed away or commoditized at the very least quite quickly. So finding that mode and then building your, at least, what could your product service trashy be around that? What is your recruitment or your set of skills that you already have in the organization? That is a, I've got it here as product-ties for your patch. But I think that's a Claudeism. But the reality, I think it's a nice line, focus on what you're good at and then build your trashy around that. I think own the orchestration and the governance layer. So the model, I've talked about here, it being done by Claude or it being done by OpenAI. Actually, ODE is supposed to be Claude first, not Claude only. Microsoft is multi-model, so they can host any of the weights and models. So the reality is that it's not really about the model. It's about understanding the harnesses that sit around it. What works best for the workflows that you're trying to tackle and then orchestrating all of that? That is your responsibility or at least that's where the opportunity is going to be. I think in terms of go to market, there's a couple of to-dos. I suppose there's a lot of people flocking to the frontier labs and their channel partner networks. I think it was OpenAI that said that they had 40,000 applications for a hundred sort of partner spots. So you can see people are trying to get the badges. I don't know if there's anything wrong with that, it doesn't hurt. But obviously, to get the badge, you're going to need the skills. So there's a lot of stuff you need to do before you can just run at the badge. I think probably Microsoft and what they're doing with this frontier company, but they're more broader frontier strategy. So our colleague Nathan, he's writing a series of notes at the moment on Microsoft's AI strategy, their recent MCAPs event, but then also the frontier strategy that they're trying to push in with partners. We'll have some research coming out that soon. But that is a natural place. If you're already an IT services company, you've already got a relationship with Microsoft. Maybe that is a way to tackle some of this stuff in an ecosystem you already understand. Because it's Microsoft, they will be channel or partner first. At least they'll say that verbally, but they want the partners to be the ones delivering their stuff in the end. And then I think to come back to your point about ROI, because it's becoming so important. I think lead with a costed use case, it really has to come down to that. Don't go into an engagement going. We can do AI for you. I think that will land very badly. Instead, we've done this kind of workflow. We know your kind of organization has this kind of workflow and we can use AI to help improve it and drive it by 10%. You've got to think like a software company. Software companies have been doing this for ages. You're an HGM SaaS company. You've gone using our software. We'll get you 5% ROI per year on your software. You've got to start thinking like that with this stuff. And then how you take that project, because this might start like a project and how you expand that into a new team. Can you support that application that you've built or can you support the underlying plumbing that you've done for that customer? That's I think where the important bit is going to be. Recruitment retention, I think it's going to be an interesting, I have no answer for what this might mean. This idea of FDs is very interesting. But how does that trickle down? How does that move out? I have some positive views on it, but it's going to be complicated to see. So it's really trying to understand within your vertical, within your subject matter expertise, do you have a set of skills and a set of people in your organization that could be the driver of your product service strategy? I think that'll be the big one. And then go talk to customers. Actually, some of the companies that have done this really well for me, they've started by actually finally going and talking to customers and going, I don't actually want to talk about technology. I want to talk about your business and what are your key issues. Sounds very basic. This is not some grand sale strategy that's new, but this is really important if you're going to figure out where to source opportunities in your existing customer base. Because that might be then the source of your new logo as well. If you can use your existing customer base, build a product service strategy around that. That might be what you use to differentiate yourself when you try to go after new customers in that same area. So a lot of things to do, a lot of interesting opportunities, but in the interest and the spirit of being a bit more positive, this could be a real opportunity, but you've got to really have the right strategy. Okay. So yeah, so I suppose the thing to finish it off with is that move by the big guys is a signal point for everyone else basically that there is an opportunity here. So step one, there is an opportunity here. Get that product service strategy right. Figure out what your differentiator is. That will allow you to have some protection against commoditization and then figure out what's right for your customers and then and deploy that way. And then that I think is a way to make this an opportunity rather than a threat. Yeah. Okay. And I think in some ways, we'll take another quick break now. But in some ways, the narrative is very similar. On the third kind of key topic we're going to talk about this show is on the SAS POC ellipse. It's a similar dynamic in the sense that it's been a wake-up call. And weirdly, that presents a real, it should underline the opportunity, yes, rather than be seen as a, as a massive risk. But we'll take a quick break now and we will dive into SAS POC ellipse next. Okay. So our third key topic for the podcast is quarter is, is as we've trailed a couple of times now, the SAS POC ellipse. But before we dive into that, just to sort of thought about where, where we're at. And there's been a lot of, it's cool. It is prompted a lot of conversation about what software companies are actually for, what software is actually for. And fundamentally, what they've been doing for the last decade or two decades really is building products. And if we really strip it back, our view increasingly is with software companies that over the next decade or so, the most if not all software companies are going to need to go in one, one of two directions. They're either going to need to deliver outcomes, so service as a software or service as software as we describe it, or they're going to need to become part of the infrastructure layer that enables AI. And we kind of believe that there's going to be less and less room in the middle. And I think that's what we're seeing already starting to see that happen to some degree. So in this section of the show, well, explore why investors do remain quite cautious about traditional SAS. What do we think an AI mode increasingly looks like? And why the SAS POC ellipse is forcing management teams to really think their business model and their products strategy from the ground up in this whole sort of concept we've talked about of a wake-up call is really where we're thinking about the SAS POC ellipse now. So that's the job of work over the next few minutes. We're just quickly recapping before we dive in about what we talk about with the SAS POC ellipse for those who have been living under a rock in our industry and haven't spotted it. But really, what has happened over the last six months kicked off earlier in the year, going back to what we talked about at the very top of the show about the technology advancements, particularly with anthropic in kind of January, February time, really put the wind up the investor community about what this meant for SAS companies in the future of their value creation. And what do we really talk about here? Features being commoditized. So particularly we're talking about things like note taking, summarization, reporting tools, dashboarding, translation, copywriting, etc. These increasingly and sort of everyone suddenly worked out that these are things that AI and certain circumstances could do cheaper, better, faster or certainly better and faster cheaper and faster, not necessarily always better, but good enough. Switching costs therefore partly related to that are falling in some areas, whereas you know, one of the key value drivers of software has been stickiness and that's being eroded or the perception is that's being eroded for some to some degree. And that workflows that were the that were really the preserve of software companies or that being managed by software companies were being automated using AI rather than software. And then the competitive advantage moves away from features as a result of that into workflow. And something we talked about extensively at our state-to-the-nation event in January and we've talked about a lot since its pricing and the whole pricing conversation, but this idea that and the kind of value driver of of NRR being 105, 110, 120 percent because of pricing drivers per seat mainly kind of goes away or potentially goes away over the medium to long term. So so investors have taken fright and they've really been looking trying to feel their way in terms of what is an AI mode. What is the what is a competitive mode in an AI world for a SaaS company? Is it proprietary data? Is it being fully embedded in complex workflows? Is it as we talked about in the services ecosystem, customer relationships, industry alignment, that distribution and customer trust? All of those things will come back to talk about those. And let's you know, this is a really big deal. This has been particularly private activity driven. I'll talk a bit in a bit about how this has impacted public company share prices, but for it's been particularly prevalent. I think in private equity world and you know it's estimated that around 25% of all private equity transactions globally have been in the software industry and that kind of 10 to 15% of AUM assets under management are in the software industry. That translates to depends who you read but anything from half a trillion to a trillion dollars of assets of cash invested in the global software industry from a private equity perspective. So and that's just PE and I don't even know what the public company number is but it's multiple obviously many times that. So there's a huge amount of capital at risk here and just to summarize this idea I talked about a second ago about this this ultimate destination for SaaS companies. Be fair this is a this is an emerging conclusion if that's can that be a thing? Anyway you know what I mean? You know this idea that really the the strategic options for software companies are bifurcating into this idea of services software this fundamental shift in business model for software companies where the natural point of the natural destination if you like is well look I'm selling products I have been selling products but ultimately I have to sell outcomes. I've gone I go from selling CRM solutions to I need to find you more customers from I'm doing HCM software to I'm onboarding a new employee or I'm offboarding a new employee or I'm finding you a new employee. So these action these actions rather than features that naturally lends itself to an ultimate destination of this business model which we don't really know what that looks like yet but it's this service has software this might be a stupid question but our software company is actually geared for that now they they used to building a product and going I will create create your fantastic applicant tracking system. If you then come to them and challenge them we go no I don't care about the product you can use what the platform you like I want you to find me or help me find my next. Well ultimately what it comes down to is price is pricing model. So the two things are absolutely two sides of the same coin. So they're not services of software does not mean service with people like it does now it means a genetic service but it's so it's more about I'm paying you I'm paying whatever sales force or HubSpot whatever per user per month fee but actually what I really want is I want I want that to find me a new customer or whatever you think that metric is it is it's a great question you ask because it is it is the big one of the biggest question unknown kind of outcomes of all of this is what does that look like we don't know that yet but so it's not it's not so much the delivery model I don't think the delivery model changes that much in the sense of it's still technology you're delivering you're not delivering people so when we say service we need to it's it's it's service as an outcome not services that need people there may be an element of that and it may be this meshing of BPO and software then certain circumstances but fundamentally it's still technology you're delivering but the out it's the outcome that you're pricing against and that's where the two sides come together and then the second kind of avenue if you like is this idea of being part of the AI plumbing do you have proprietary data fundamentally and interestingly I think that's where software this is where software and information services kind of co-less in some ways but also workflow sticking to say you know you service now for example you know absolutely embedded in the plumbing yes they do a whole bunch of using that example of task workflow related stuff but actually you know one would argue that actually the real value of something like service now is in the plumbing and connectivity so you know you kind of the middle ground is system of record undifferentiated system of record that is has doesn't that the whole GRR stuff we talked about earlier and that's not a great place to be so from just sort of pull this strand together understanding AI moat I think as in competitive moat in an AI world will I think and having a really forensic lens on that as a company will help software companies decide which way they go understanding if you for example if you really think that your proprietary data is your moat it probably lends you into it but but your features are quite commoditized so you might be an assessment software company in the HCM space and actually assessment the software and the product than the actual functionality around assessment is is not complicated you know you could vibe code that aspect of it but actually the proprietary data for what good looks like from an assessment perspective is extremely valuable so that might lead you into thinking actually well I'm not a product here I'm not I'm not a desktop product but I am going to massively improve outcomes for assessment so you sort of mean that's interesting I suppose because I've already you know you hear stuff on the wind but you talk to people and says okay well I'm I'm really powerful because I own all that you know my my system has all the customer records or the HR records so I'm I'm pretty safe because no one's going to be able to replicate that which is true my worry is if they lean into that and just go well I don't need to bother with any of the features if if the system of action is going to be really competitive market and I'm quite safe as a system of record I won't bother but I think my view is does that not then the system of record will become come on like at least the pricing that you can get out of a customer you're not bothering to compete in that space yeah you won't go bus no no no saying that but actually just a commoditization you know grow you know valuations in software are partly predicated on on recurring revenue and the kind of the solidity of the business which which arguably doesn't go away in that scenario as you say because you remain a system of record yes yeah but the growth goes yeah and and arguably you go into modest decline and so there's a big difference I think between the SaaS era and the the current era or what we're entering now is that in a SaaS in the SaaS era those that stayed on prem because the stickiness was still there when the move from on prem to SaaS you you continue to probably you still had stickiness and and the cost of changing was still very high actually you could still probably put inflation rises through so your GRRs floated around 100% and you may be got a new customer here there so you could probably still grow it made even high single digits that doesn't that doesn't that doesn't really work in in this world so the the risks are all the potential outcomes are more significant so and actually I think it's worth when we think about how this plays out it's actually worse spending I mentioned the HM software space there and I think it's kind of interesting to think about that as a because what what is actually what what what's happening with SaaS apocalypse so we I think we sort of gone over what it what it is what what is actually happening and actually what is we are now seeing which I think is super interesting is we are seeing growth rates changing as a result of AI adoption in customer basis and HCM is really interesting when sort of sat down with Cameron R are leading this part of the market and megabyte and it's really clear so the HM so in the megabyte universe the HM segment has gone from being one of the fastest growing segments that we track at megabyte in the software space so we're looking at the growth rates of UK headquartered HCM software company this is HR tech HR tech HCM human capital management thank you yeah into the one of the slowest so there's been a there's been a really tangible deceleration in that and that's and that's it's not all about AI talk to Cameron about it he talks about the fact there is a commoditization commoditization there but those two again use the same kind of the same two sides of the same coin yeah it's commoditizing because the growth drivers are in AI and and building that I also think with HCM there's an there's an an EDO sync and not a very pleasant EDO sync see for vendors there you've got this AI this sort of commoditization going on and you've got AI in some areas particularly around the the sort of workflow area so you know you can and a sickness reporting holiday requests all this stuff is very easily I wouldn't say vibe coded but it's not it used to be a feature that you could upsell and now it's it's not really anymore but at the same time you've got no real customer pull from HCM because they're all terrified of using AI because the data they have in H in the HR department is super sensitive right salary data bonus data et cetera so you don't want to you just like you're in a status because like which is probably right yeah so there's this it's the worst of both worlds so I think that's we're seeing tense the point I really want to get across we're seeing tangible impact from this on the on the growth rates and valuations so we look at the meritech this is a bit like the best of adventure partners it's a it's a company that that assesses SaaS valuations very effectively and recommend listeners have a look at it it's really useful if you're interested in that kind of thing and you know probably don't need me to tell you that the software stock prices have been hammered over a lot of six months of results this the median I think it's probably got 100 companies in the SaaS index that they run at meritech down or the median reduction in share price to 21% but lots of companies like HubSpot and others 40% plus down year to date and the median EV ARR now is three and a half times and the median EV EBITDA is 13 and a half times next 12 months so that's come way down over the last six months HubSpot's trading on two and a half times EV revenue work days trading on three times just over three times sales force trading on three point six times I mean these are companies that have traded well into double digits revenue multiples in the not too distant past so this is a this is a big deal for share prices and valuations going back to the HCM example and HCM software companies again this is a global view they are the median next 12 months trading multiple of EBITDA is 8.5 times now these are HCM focus software coming so zeros in there the accounting software but they do a lot of HCM stuff as well is the highest at 17 times so that's topping out now that that peer group at 17 times EBITDA pretty striking and when we think about the impact from a deals perspective it's interesting because in prep for the show I was looking at the deals from a looking at okay what we do this every quarter obviously and it's in the barometer reports that we publish for subscribers can look at those for one for ICT services and one for software. The private equity deal flow in software, actually, interestingly, because I think we said last time on the podcast and probably the time before, certainly last time, that we expected software deals to, like, private equity investors are very cautious. And actually, I was looking at the deals going, actually, there's 29 private equity deals across software and ICT services, about half of those are software, which is kind of the same as the previous quarter and not too different from what we've seen previously. So I thought, that's interesting and I drilled, but I drilled into it and I was again chatting with a team of megabytes about it. And what you see is that actually, over half, between half and 2/3 of those deals are not SAS deals, they're actually information services deals. So we define software and megabyte as software and digital services, including that is information services. So Pee have decided that the best information services companies that've got proprietary data and information, and that is a better bet for their money than SAS at the moment. So optically, it looks like there's been no impact when you drill underneath it, the actual number of genuine sort of traditional SAS deals has dropped significantly over the last three to six months. What's the play with those information services that use the IP and then what build feature set on top of that? It's into that kind of bug bifurcation, not bifurcation other way around, the sort of combination of proprietary data, whether you're a software company or whether you're an information services company. If you're getting into that information or data or both is embedded into workflows of the client base and an AI level, that becomes incredibly sticky and valuable. So that's, I think, at a very high level of thesis. From an M&A perspective, you've got, sorry, my favorite word today, bifurcation, you've got two things going on. You've got the SAS 1.0 pre-AI roll-up stuff still goes on. But the valuations there have dropped dramatically. The clearing price now for a small SAS company from an EV with our perspective is high, even mid-single digit EV dial multiples, whereas it was 50% higher than that, 12 months ago. Ben, you've still got a lot of the cows in for share, true layer, for taros, going and all doing deals. Although interestingly, access has an access group. As you know, one of the biggest acquires hasn't done deals since February, which I think is really interesting. So it's been quite-- R has been relatively quiet as well. I think the HG portfolio, I think, is a feature here because they are all over this. And there is also some expensive AI, not as much, but there's some AI expensive. So duck creek buying send, sage buying doyen.ai, these are more expensive, very clearly AI-focused. That's what I mean by the bifurcation, got it? Volume continuing, but at a much lower price, AI-focused product development, sort of, by rather than make AI-product. It's those two things. If you're doing expensive SAS roll-up, relatively, you're definitely in the wrong place. I can't wait for all the companies to start rebranding themselves to have AI in the name or to thechangerwebsite.air. And hope that that gets you in the game. Such a bad idea, such a bad idea. It was same with the dotcom, exactly the same. Everyone was like, we're dotcom. Oh, no, actually, we're not. And then just I couldn't go on without giving us a plug to talk about M&A at one of our events. In September, 29th September, we have the co-CEO John Jorgensen of Access, as in one of the speakers at that event. So subscribers, actually, I think we're standing alone already. But if you haven't, we're looking at this whole idea of this bifurcation of strategies in 29th September. So if you're a subscriber to megabyte, you get access to our events as part of that. Go and have a look on our website and you can see what's available there and you can register. OK, so I think I just wanted to talk about the response here before we wrap up. And I think I wanted to mainly think about the response this and what-- because I talked to the top of this section about how the best companies are responding to this as a challenge to be overcome. And frankly, as an opportunity, we sounds a bit counterintuitive potentially. But actually, if you're a software company, if you're an established software company, what are you doing about this to see this as an opportunity? And there's one-- I won't mention the name, because there's some sensitive things in here. But there's a software company we know very well. It is, I think about 200 million of EBITDA, to a decent-sized, larger mid-market software company, coming up to the reason I didn't want to mention the name, because it's coming up to its probably next event. It's been in the portfolio of its-- if it's-- tonight's G, actually, if it's been in the portfolio of its investor for a three or four years now. So I sat down with this purely coincidentally with the management team, CEO, and CFR of this company, just for a lunch. Just as all this-- the ship was hitting the fan with the SaaS apocalypse earlier in the year. And they'd been talking to private sector back in the last year about-- and they were knocked over in the rush, because it's a really well-run company, really solid. Probably the technology, because of the markets they're serving is probably not as advanced as others, but that's fine because of the markets they're serving. And literally, this SaaS apocalypse happened, and then it was tumbleweed. No one was returning their calls from private sector, and they were really like-- when I sat down with them, they were like, yeah, we were just trying to work out what they were doing. It is, by the way, I think, one of the best management teams I know in this market. As it happens, we had to catch up with them on a call last week. And I was absolutely blown away with what they've done since we last met. They have got a fully worked-up internal AI implementation plan. They, by their own admission, are still working on their product structure. I think that's a feature here. They've identified 7% of EBITDA cost savings that they're going to release over the next 18 months. And they fully re-engaged with private activities. So the best management team-- the key takeaway here is the best management teams are on it with this. And it has been-- we're using that term again. It's been a wake-up call for them. And they are seeing it as an opportunity. In this case, initially, they are going to see-- this is a margin enhancement, but actually, they are all over it. And I'm sure they'll get their product strategy right as well. So that's the key thing. What does that look like when we're trying to start thinking about how we pull this all together? Thinking about-- we've talked about this so many times today. What is my mode? How do I remain relevant to my customers? Because it's kind of existential for this, as it is in services. How am I driving values to this pricing conversation? How am I going to drive value for my products? Emily, I do have to start thinking about this as a kind of pricing value-- value creation, not in the broader corporate sense in the very much the product sense. How am I going to get value for what I'm delivering to my customers? What is the right organizational design? Because that is becoming kind of critical. Difficult conversation, but how many people do I need in my organization? And that is something I'll come back to that. And is my funding right? A lot of these companies are very highly levered. And we haven't talked about it a lot last time. We haven't talked about that here. I talked a lot, just as you have a drink of water. Any perspectives on this before we start wrapping up? I suppose whilst it's about SaaS and it's about the software companies, I think this is all just as relevant for ICT. Some of the dynamics might be slightly different because software is ICT. ICT never benefited from the absolute highs of software valuations. They had their own highs in the 2020s, 2021. But the reality is all the questions that you mentioned are still exactly the same as I talked about for the second topic, which is the mode, the subject matter expertise. We might use different words for it, but it basically means the same thing. I think the thing that sends that sound up to me is relevance. So this is kind of what we're going back to with a lot of this stuff which sounds really basic, but is your relevance for your customer? The same getting better, getting worse and that's simple. Even just a red amber green around that, I think should be something that's just path of the course with all this stuff. It's just to really understand the stuff that you're doing, is that the stuff that your customers will need in five years' time? Is this a get ahead of the conversation? Even about pricing you've talked about in the past, right? We know there's a couple of different pricing models that everyone's trying to throw around. Maybe some people don't really have an answer around that. But the reality is you should have the strategy so that when the conversation comes, you're ready for it. And a lot of this stuff is quite the same. But I'd say with product services now, you need to be building already or at least you need to be building or plan already, which is I think the exact same thing in services. Yeah, I think that's great. And so how do we summarize this? So the SaaS box lips really is, we use the term a lot. There's been a wake-up call for software company CEOs really has been, but critically, the best of them are rising to the challenge in impressive fashion. The short-term priority is nailing operational optimization. I think that's clear. But very quickly, I think this needs to turn into a ground-up review of product strategy. The next stage will be to deliver real long-term, shareholder value will be ultimately lead all SaaS companies that we talked about this at the beginning, to one of two distinct paths. Services of software or AI infrastructure. I'm not pretending that that isn't long-term because it is a long-term thing. But thinking like that, as you said a second ago about thinking about the five years fundamentally, how does that translate to? OK, what do I actually do if I'm going to take this back? I've listened to the pod and I'm going to take this back to my management team. At a high-level shift gear. And I think that's something we've seen the best CEOs and the best boards do since SaaS poplits happened. Thinking about this with greater pace, greater intensity, and more fundamentally about what this means for your business and we've talked about all those things, the top-quartile SaaS companies are reacting fast and their pace has fundamentally changed. Ask the difficult questions, secondly. And in many cases, I think we are sadly going to see headcount reductions. We might not see a lot of it in the press because it's going to be feedback businesses or owner-managed businesses that aren't certainly going to announce this stuff. But I'll give you another anecdote. I was chatting to the CTO, actually Chief Product Officer of a decent-sized software company the other day that has 350 developers and quite a lot of them are offshore. And he was saying, look, we have pods of 10 people in our development teams. We are going to move to a structure that is pods of 3 or 4 and the rest is going to be agentic. And we are going to probably lose 100 of our 300 development heads in the next 18 months. A lot of that will be offshore and outsourced, but that is the reality of this is that I think we're gonna see that. Others have talked about their service desk are now in service as you talk about this then. We haven't talked about head camera reductions, but efficiencies in service desk will often mean your heads. So just asking the difficult questions in the business to get to wrestle ROI from this. Clarify your pricing options. And we still don't think it megabyte. I don't think it's true in services. It's certainly not true in software. What we know, we don't think I know what the next thing looks like, but understanding what you could do. We say there's a lot in our presentations. Understand what you could do. Understand what the impact that is gonna have if you do that on your profitability in your business model. And then experiment with what that looks like with perhaps a very friendly core customer or a prospect situation to ever see how that works. And fundamentally, sort of finally, build ground up, build your product strategy when we think about that from the ground up. We talked about this many times on this show this time. AI is not a feature. Product strategy response, I think, will be slower. And that's probably okay in most instances, particularly if you're in markets and not adopting that quickly. And you can move through this kind of continue, we talked about a megabyte assistive AI, partially agentic, fully agentic over time. But you need to understand what that looks like and you need to think fundamentally about that. And then just to conclude, again, about the two kind of ultimate destinations, one of which you'll need to pick. Software, services as a software or AI infrastructure. So I think that's all I wanted to say. Anything you wanted to just add before we wrap up? So those people are saying we're quite negative. We were a bit more positive this time. We were a bit more, we're trying to be, and actually just to end on that note, I think that there are risks in AI. There are risks in the services ecosystem that we identified in the second, second of what we talked about. But we really do genuinely see this as an opportunity. The AI Goldilocks period is all about grabbing the opportunity now in a set, insensibles are on word. Grabbing the opportunity now in a structured, focused way so that during the 2030s as this stuff really becomes mainstream, you are best positioned. That's fundamentally the opportunity is there. I think the SaaS apocalypse and the frontier labs going into services are two wake up calls that should be prompting CEOs of mid-market software and ICT services companies to grasp the nettle. Thanks for listening to the CEO, Bronner. If you found value in this episode, you can explore more expert insight, sector analysis, and proprietary intelligence that sets us apart at megabyte.com. Follow us on your preferred podcast platform and join us next time for another conversation shaping the future of UK tech.