From SaaS to CaaS —Fireside Chat with Scott Brinker & Erik Charles

Julia Nimchinski:
Awesome! We now make a move from SaaS to CAS. Eric Charles, Fractional CMO and CRO, leads this conversation. He’s joined by the godfather of Martech himself, Scott Brinker. What a treat! How are you doing, and what’s in your agenda Coas?

Erik Charles:
I mean, it’s a wild world. I’ve actually started playing around with building my own agents for a while. I will admit to definitely messing up a few codes, a few keys. Luckily, I keep it all in the sandbox. I’ve yet to release any of my creations outside of the sandbox so far.

Julia Nimchinski:
How about yourself, Scott?

Scott Brinker:
It’s all agents all the time, where I am, so.

Julia Nimchinski:
Awesome. Eric, take it away.

Erik Charles:
All right, thank you, and this is… this is great. I love… I love, Julia’s great at coming up, you know, we’ve got the godfather of Martech, and I’m apparently the CMO whisperer. Although, coming out of product marketing originally, I used to be… call myself the nerd whisperer. I always had to show my skills by saying that I programmed in Lispin Prologue, and then sometimes the engineers would listen to me for a few minutes.

Scott Brinker:
Yes, I get that. As a recovering engineer, I get both of these.

Erik Charles:
There we go, you’re there as well. I read a great quote the other day that COBOL is the asbestos of the programming world. It’s really effective, really dangerous, and scary to remove. So… But let’s move closer into the future. You know, I am happy to be joining with Scott Brinker. He’s talked about the context as gravity, he’s talked about the shift from software as a service to context as a service, SaaS to CAS, which I love and embrace that concept. You know, and it’s, you know, how can companies or people, you know, we say organizations, but it’s really the people leading those organizations, get around the Chaos, I’m gonna call it chaos, chaos as a service, in the current AI-driven world. I think we’ve got a lot of that. So, Scott, thanks for joining in for, what have we got, 27 minutes of talking about this.

Scott Brinker:
Sounds great. I love Chaos as a Service. That is the most accurate description yet. Chaos on demand.

Erik Charles:
I think it… I really do think it is. So, but let’s start with the first one. Okay, you came out and played the SAS is Dead card, and I’ll admit, you know, I’ve heard that around a couple times, but then I go, and certainly I look at my investment portfolio, and it might not be dead, but certainly it’s hurting and bleeding out on the tarmac right now, which is not fun. You know, so, take me through this. Why is context win over, software?

Scott Brinker:
Yeah, I mean, actually, I’d like to think I was trying to mount a bit of a defense, for the SaaS industry overall. I mean, I think, just to level set, I mean, obviously. all software is going through an incredible period of disruption, a ton of uncertainty, so I think even what you see in the market isn’t necessarily a reflection of a bet that, oh, these large SaaS companies aren’t going to survive. I think it’s more of just a… a pricing in a new level of uncertainty of, like, you know. for the next 10 years, we’re not quite sure what this is going to look like. So given that, I think it is absolutely imperative for SaaS companies to figure out how they make the shift from what made them so strong in this previous era we’ve been coming from in SaaS general, you know, my expertise is really on the, you know, marketing and sales tech side of stuff, so certainly in MarTech, there’s been a very well-defined generation, you know, of leaders in that space. But as we start to shift into this new environment, one of the things then I think I would frame it this way. A lot of SaaS at scale in the previous era had been about consolidation, right? It’s like, okay, there’s gonna be these many companies offering CRM, or marketing automation, or whatnot in the space, and they’re gonna duke it out, and acquisitions, and just, you know, battles, you know…

Erik Charles:
I don’t need single sign-on, there’s just one application.

Scott Brinker:
Exactly! Well, and you know, that’s basically what a lot of the strategy was. Like, okay, let’s keep narrowing down the competition, let’s pull in adjacent categories, try and build these things as sweet-like as possible, and to be honest, that actually caused these companies to scale. But one of the assumptions in there was, okay, this thing about building software and managing software and running it, right, that it is to everyone’s benefit to that to compress. And I think one of the things we’re realizing with AI is, okay, whoa, these days of somehow creating software being scarce. You know, that somehow scarcity is, you know, the function. It’s just… it’s wrong. I mean, people are just going to be building more and more software. I don’t think it’s all going to be vibe-coded, although there’ll certainly be a lot of custom apps and agents that businesses build on their own. I also think it’s likely you’re going to see a lot more people bringing commercial software products to market, because, hey, if I’ve got a great idea, and you know, there’s a niche I want to go after, I can build it, I can get it out there. So, that’s what sort of led me to think that, like, okay, well, for these large SaaS platforms, just trying to compete on a path for consolidating all of that. I think that’s a tough hill to climb, but… The flip side of that is if you really do think we’re gonna go through this shift from scarcity to abundance of software. then there actually are a lot of real problems with the abundance of software. There’s like, okay, well, wait a second, you know, how do we keep this stuff coordinated? You know, how do we manage the data? There’s governance, there’s compliance issues, you know, and honestly, there aren’t really great solutions for that today. We’ve sort of, you know, the explosion has happened in advance of, you know, a really clear framework of, okay, how do we We manage this at scale. And that’s what I think is the opportunity for these platforms that really can serve, I call it context as a service, you know, because it’s not just about the software features they have in their own product, but it’s the ability to support the context for any other app or agent that businesses want to either build or buy that are gonna need some sort of organizing framework.

Erik Charles:
So, so, I gotta… it’s a… so I, you know, when I started in SaaS, of course, you know, and we’re sitting there. and built it out, and then I remembered I went through the process of APIs, and had to learn what a REST API was, and how do we get this to connect to everything, you know, from FTP sites, and I know there are companies that are still sharing data. through secure FTP. FTP hasn’t gone away since I was downloading questionable materials off of an IBM AT back in the mid-’80s. But then we went to… so then all of a sudden, I agree, you had consolidation or hyper-integration, so you had all these little, I call them, you know, data ponds sitting there. Then companies started claiming they had built data lakes. No, no, we’ve got all the data together so that any of our analytical tools, which is an interest of mine, could see the full picture, but… I felt that data lake was one of the great lies, along with seamless integration. Because it still felt like a bunch of ponds with maybe little channels between them to try and keep it going. It might be slightly more secure, maybe having a master identification number for be it an employee, a position, or a unit that’s there. And so on… in the AI, in the Gentic AI world, as we know, one of the biggest challenges is, is it being fed good, clean data? So where… does data end and context begin? Give me the VIN diagram of your context as a service, because I honestly think sometimes DAS, of data as a service, this is just software that’s using all this data, because everything we do in the business world is data-related.

Scott Brinker:
Yeah. Alright, so there’s maybe two parts to that. The first part is this move towards cloud data warehouses and lighthouses, which has absolutely been a movement. I’m actually a little bit more bullish on it, mostly because I feel like the number one problem we had in MarTech, and I’m sure this extends to these other categories as well, too. is the integration problem in the sense that we just couldn’t get the data, you know? And the process of trying to get the data was just… I mean, we’ve spent two decades trying to do it, and still people list it as their number one problem. You know, the advantage, I think, of moving to these ideas of a universal data layer is, like, okay, if everything can push data down into it, and everything has the ability to pull data out of it, then at least at a fundamental, like, data flow level. you sort of get past that first problem. Now, that being said, the moment you get past that first problem, you immediately then discover the second constraint of, like, oh, crap. you know, my problem before was I couldn’t get all the data. Now the problem is, I’ve got all the data, how the hell do I make sense of it? You know, and this leads very nicely into… the reason I framed this shift from being from SaaS to context as a service is really, riffing off of the idea of context engineering, you know, that has become sort of the… the label du jour for, you know, what we’re doing with a lot of AI and agents, where it started back with prompt engineering, right? Like, just, oh, how do we feed the right instructions, you know, to this AI engine to get what we want? Context engineering, my oversimplified way of thinking of it, is it’s expanding on that to say, well, it’s not just about giving, you know, an AI assistant or an agent instructions, you also want to package up the relevant data for it, you want to make the relevant tools accessible to it. Because if it tried to, like, on every single instance of doing something. Try and figure out, from first principles. all of your data, and like, what… I mean, it’s just, right, it’s just not a, economically or, like, feasible way to do it. Right. You know, so what people have been doing is when you’re custom building this, and you’ve got professional engineers working on this, context engineering is like, okay, well, how do we find the right data, package the right data, get the right tools? And it works. It just turns out it’s a fairly complicated and technically challenging thing to do when you’re doing it manually. And so, my hypothesis here was like, okay, well, wow, we really are going to want that sort of context engineering functionality delivered to every app and agent that we’re running, you know? Isn’t this an opportunity for these SaaS platforms to really understand the domains in which they operate, you know, in marketing or sales, you know, to serve in that role? of packaging up the data, whether they’ve got it natively or pulling it from, you know, like, say, a data lake or, you know, data warehouse substrate, being able to package that up, have the relevant tools, if there are certain key decisions that it might be making to optimize, not just for that agent, but sort of across the entire business process. Package that up, deliver it for apps and agents to leverage. There’s real value there.

Erik Charles:
Now, is that packaging up what you mean? You know, we were talking about this when we were preparing for this little bit. I always love when our prep calls take longer than the actual, you know, session, but that’s all right, of decision architecture. So, is that what you’re taking this to? Is companies have got to do this from a decision architecture perspective.

Scott Brinker:
Yeah, I mean, part of the reasons we’ve been having a lot of discussions about decision architecture lately is, you know, we want to, you know, give these agents the ability to actually make decisions on their own, you know, have greater degrees of autonomy. But that’s actually raised a question for a lot of scenarios of, like, well, wait a second. how do we make decisions about this? Because it turns out, particularly in the marketing domain, you know, a lot of the way decisions have been made on everything from, like, okay, exactly how are we defining our ICP for this campaign, and, you know, why do we do this budget here or there, or, you know, how do we align these messages? is incredibly tribal, it’s done through, you know, things that people are having discussions, you know, maybe in Slack, maybe on a whiteboard, and things that aren’t, you know, it’s all this stuff that, I mean, you know, I can’t pick on it too much. It’s worked for marketing, but it’s not been something that you’ve been able to actually capture, like, okay, this is actually the decision logic and the decision guardrails. In the decision process, so that if we wanted to turn over the ability for agents to start to take on some of that decisioning. that they’d have a framework to operate in. And so I think that’s sort of the first step, is like, I don’t think software’s gonna do this automatically for us. Companies have to start to invest in actually defining, you know, what their decision structure, their decision architecture is. But as they decide that, I think this is, again, a role for these platforms that are acting as the coordinating layer inside the tech stack to be the repository where that decision framework is embedded, so it can provide that governance to the apps and agents that run on top of it.

Erik Charles:
Well, and this is where it gets interesting, because it’s like, we’re conducting field experiments here. You know, we’re, you know, in the sense that we think this is right, and I’ve seen this as well. I’ve seen the problem of the recency effect. Where everybody says, no, no, no, the ICP is this, because that was the last 3 deals we closed. Yeah, but we closed 175 deals last year. what did they all look like? And if we backwards apply our… just our… our mapping of ICP, we think ICP is this many employees, this… this industry, this location, this… these are the other tools. If we took that to grade all of our leads by A, B, C, D, and applied it to all of our one deals last year, how many deals did we win that were A, B, C, or D? And I’ve done that with multiple clients of mine, and Every single time I uncover… so you’re gonna throw away 25% of last year’s business because you think, by your new definition, they are bad deals. And you can just get these. reactions, because I’m challenging the tribe at that point, which I guess is why, you know, I come in as a fractional, because they cut me into pieces by the end of the project. But it’s necessary to ask that question, so… I mean, so, you’re saying… So is the platform going to own the decision? Is it gonna be the rules of the road? And how often do we change the speed limit or the exits? You know, what’s the mod… so it doesn’t… because we don’t want to modify it too often. You know, how are we gonna adjust those rules over time?

Scott Brinker:
Yeah, okay, so there’s a lot in here for.

Erik Charles:
Yeah, I know, I know, I have the joy. I could ask a one-sentence question, and you’re like, oh my gosh.

Scott Brinker:
I thought this was gonna be a multiple choice quiz.

Erik Charles:
Oh, man.

Scott Brinker:
Let’s see. All of the above.

Erik Charles:
I did bring a blue book, it doesn’t show up on the camera, there we go, now it catches the cover in case we need to just fill that out.

Scott Brinker:
Nice, nice, nice. Alright, well, I mean, again, I think, first thing I’d have to just acknowledge here is this idea I put forth of, you know, SaaS platforms becoming more context-as-a-service platforms. is almost entirely fiction today. You know, you could maybe argue some of what Salesforce is trying to do with Agent Force, you know, some of the stuff I see HubSpot just starting to do and trying to open up, you know, their Breeze platform. Like, you see them starting to nod to this idea of, like, okay, we actually do have this context, but it’s embedded in our applications itself. how do we start to think about being able to make that more accessible, you know, for other apps and agents to be able to leverage it? But because it’s very early, there’s just still a ton of work to be done, and I think the governance one is the largest, because The way in which a lot of governance was implemented inside MarTech tools was it was is actually implemented in the UI of that specific app. It was like, okay, well, we will decide through the UI what you can do and what you can’t do, and here’s where we will choose to put the, you know, permissioning boundaries, and which, you know, group are you participating in.

Erik Charles:
Even what’s in the drop-down. Exactly.

Scott Brinker:
Excuse me.

Erik Charles:
I’m sorry, I can’t figure out where to classify you in this drop-down list of how to code this particular, you know, client.

Scott Brinker:
Exactly, and so a lot of that governance was somehow, like, entangled with the UI. And again, not very well… like, trying to pull out the governance framework from the UI, actually fairly hard to do, because it just wasn’t how it was built. And so this is part of the process we need to do. If we’re going to say, like, okay, well, the UI for our particular application. May not be going away, it may still be useful for a bunch of people in a bunch of scenarios. But it’s likely there are also going to be other ways, including a bunch of these autonomous AI agents that are going to be wanting to interact with this, and how do we surface up governance in a way that isn’t a UI-based constraint, but it’s something that, okay, this is the instruction rules, this is how we’re going to, you know, intentionally limit the data that it can access, or the tools that it can access. So I think it’s still very early days for that, but That work is what I think these SaaS to Cast platforms should do. This is why I think they have the opportunity to have something really valuable, because I think it’s a lot of friggin’ work to do that and get it right.

Erik Charles:
Well, yeah, but, you know, but I love UI. UI is a big thing of mine, because too many UIs are either, I’m sorry, built by engineers, or built by… or built as they went along, and so you keep on adding bits and pieces to it until all of a sudden it’s a cacophony, and… I can’t even figure out how to… I was just trying to… I honestly was just trying to update a deck. And the UI for just doing fill with the color shifted depending upon where I clicked in a single object. I mean, it’s just one of those things, come on! Now, what I want to do is say into a microphone, say, fill that block with my clients. color of burnt umber, you know, from the old Crayola pack. So, are you… you know, if the UI is, you know, arguably going away through, like, natural language interaction, like, I can recognize a sales force. I can see the difference between numbers. you know, Excel and Sheets. I mean, I glance at it, I’m like, nope, it’s not here. And if you’re not copying Excel’s interface, I don’t know how to use your spreadsheet clone, I’ll freely admit it. I get angry if you don’t have the 7 features that I use constantly in said spreadsheet. So, how’s a platform gonna maintain their brand? I mean, the value, I think they’re gonna chase the value by what they return. But I’m gonna back up on a pure marketing perspective. If everybody’s just letting me talk into the microphone, and vibe interacting. How do you keep your brand going?

Scott Brinker:
Yeah, alright, so that’s a great question, and again, multiple levels here. One thing I would say is, I think if you really want to be a platform company. And not every software company has to be a platform company. I think you can have, like, the best friggin’ app for, I don’t know, say, running online events. You don’t need to be a context as a service platform. Yes. But if you want to be a context as a service platform that’s going to serve all of these other apps and agents. you really have to look at this through the lens of becoming an infrastructure company, you know? I mean, and we have examples of this, I mean, like, you know, we’re all using AWS every day. And none of us have any idea of, like, any UI access there whatsoever, which is perfectly fine, you know? Now, so that’s one level. I think you can build companies that have tremendous value that are a little bit more of an infrastructure play than just a UI play. The second thing I’d say, though, is I don’t believe that, like, UI is no longer important, and it’s not that I don’t… I don’t believe that, like, everything is going to be done through a conversational interface. Rather, I just would say that, again, if you’re in that context-as-a-service platform. The truth is, there are going to be multiple ways in which people are going to want to interact with you. You might have your own UI that, for certain use cases and for certain users, is awesome and amazing and hopefully differentiating. But it may also be the case that, you know, using things like MCP, someone might be, well, for this particular use case, I actually want to work in sheets, you know, as an overlay on top of your product, because that’s how it works best for me. Okay, great, if I enable that use case, if that helps you get more value out of my platform, sure. Will there also be conversational interfaces? You know, will there be an agent that… you know, it’s not even really fair to say it’s like a conversational interface, but a way in which an autonomous agent might want to be able to access certain capabilities in your platform and leverage in what it’s doing. It doesn’t need a UI, it needs… what would you call that? An AI? Agent Interface? Alright, scratch that. That’s going to be an overloaded, acronym. AI something. You know, and so I think there’s just going to be, like, multiple UIs, and the fact is, just like… With code. I think the same thing is coming for UI here, where… I say this with great respect for designers as much as I have great respect for engineers, just AI is reducing these things that used to take weeks or months to craft. to actually be things that people can craft pretty amazing things if they know what they want, how to do it well, like, in hours or days. And so I think you are just going to see a lot more bespoke UIs where, like, okay, for this. aim for this particular work they’re doing, they might instantiate a particular UI that really serves their exact experience for that. But again, what’s feeding that AI? Feeding that UI? What’s constraining it? What’s, you know, making sure that it’s consistent and reliable? I would humbly propose that’s what the context as a service.

Erik Charles:
context.

Scott Brinker:
enabling.

Erik Charles:
So, this is gonna be fun. So, you brought up MCP, which, of course, Master Control Program, Tron, the original, you know, computer show of the online world. I just had to recognize that.

Scott Brinker:
I’m digging all these rounds.

Erik Charles:
I mean, and then I know in one of your blogs, you talk about Tank from the Matrix, where you upload skills into agents, so I want to take the two of those combined with this issue of trust, because I think you said none of us really care about AWS in the sense that… I mean, I’m sure I’m using multiple apps on AWS don’t even notice. But I also remember the beginning of the SaaS world, you know, especially when it was originally not even called SaaS, it was ASP, Application Service Providers. And we had to get people to trust us. to take their data, and that the application would be available to them when they needed it, and it wasn’t resident on their computer. That… I mean, that was a… that was a big trust shift. That was the launch of Salesforce. How did they do that? And did the same thing at the launch of a couple other companies that I work with, at exactly, that was an issue. We had to convince people, no, no, no, no, we can take your commission data as well. You’re already trusting your… your customer relationship data on Salesforce, trust us to take your financial data to calculate commissions. Now, nobody even questions, we just assume SaaS all the time. So, how do… how does… how should a CAS platform pitch itself To be a trusted operator.

Scott Brinker:
Yeah, well, I mean, I feel like my answer to everything these days is like, well, that’s what a CAS platform can do. Right. You know, to give credit where credit is due, it was actually, from Theory Ventures, Tomas Tungus, who writes a wonderful newsletter. who actually used the example of, like, oh, wow, you know, given all these skills files, you know, that I can just, like, oh, download a skill to do this thing in Excel, and this thing, you know, oh, I want this for PowerPoint rendering, you know, that, he was the one who did that analogy of, like, tank, like, you know, and, oh, now I know Kung Fu. But he also was the one who raised the problem of, like, well, when you’re just going to these public GitHub repositories to get these skills, you just don’t know what you’re getting, and, like, there’s been some analysis, you know, more than a few of these, having, like. malware in them, you know, but the way you solve that is, again, if you have a platform that is acting as really the moderator of its particular ecosystem. I mean, I… again, I think of, like, AppExchange, you know, for Salesforce, or, you know, like, the HubSpot app marketplace, where, like, when things get submitted to those marketplaces, they go through reviews.

Erik Charles:
They get tested.

Scott Brinker:
checked, you know, the company itself, and as a result.

Erik Charles:
Repeatedly reviewed and security checked, I will note, having had to answer.

Scott Brinker:
Believe me, I’ve been through the process with both of those.

Erik Charles:
Yeah.

Scott Brinker:
On both sides of it. But, you know, the benefit of that is then customers can have a relatively good sense of trust of, like, okay, this probably doesn’t have malware in it. And I think when you start thinking that there’s going to be an explosion of more apps, more agents, more skills, I mean, what is a skill? Is it a software thing itself? Who knows? But there’s gonna be an explosion of more and more of this. The value in having, like, a trusted ecosystem becomes increasingly greater based on just the sheer volume of things that are out there. So again, if I’m a… if I’m a SaaS platform that’s trying to get to the next generation, what is my next life, you know, in this AI world? I think, wow. As a context as a service platform, I can take the responsibility of getting the right data, the right instructions, the right tools, you know, helping the right decisions, you know, a good governance framework for everything happening in my domain. Oh, and yeah, if people want to be leveraging skills on top of this, great, we can have a process by which we vet, you know, ecosystem skills that, you know, they’re validated before they’re made available to anyone. on our platform. There’s a whole bunch of these things that I think would be incredibly valuable, you know, to customers. It looks a little bit different than what the SaaS platforms of this past decade have, but let’s face it, things are going to look different. The question is, yeah, what becomes the vision for this next generation?

Erik Charles:
Okay, quick lightning round watching the clock before Julia comes after us. One, we’ve gone from, here’s the application cost, to here’s the per-seat cost. And some people are playing around with consumption-based pricing, which honestly gives me flashbacks to cell phone minutes, so I’ve got a feeling that’s under threat, or maybe that’s my bias. What’s gonna be the pricing model for Contacts as a Service?

Scott Brinker:
Yeah, I think it ultimately is an infrastructure play, and so I think it will be largely consumption-based, and I fully acknowledge that is a transition, both for the vendors and for the buyers. But I think we’ve also got a window in which we can do that, right? You see a lot of SaaS companies that are doing that now, where they’re like, okay, we’re starting to introduce consumption pricing, but we’re going to include a bunch of consumption in the existing, you know, per seat structure. You know, so for most customers, they can get used to that, they can follow it. And the truth is, like, for people who are in IT and software development, we’ve already become completely adapted to consumption pricing, again, off of, like, AWS and Azure and GCP and all these sorts of things, and it turns out, actually, once you’re doing this, and you develop the models, it is relatively predictable on that. So, I know this is new for a lot of other departments, marketing included, but I feel like we can have a decent transition, and I think the end state that we will get to will actually tie value much more closely to what we’re paying.

Erik Charles:
Okay. Well, thank you very much. This has been a blast, and I’ll let Julie come back in, because I can see we’ve got the next panel getting ready to speak.

Scott Brinker:
Eric, thank you so much.

Erik Charles:
No, this is fun, Scott. We’ll have to catch up with each other soon.

Julia Nimchinski:
Thank you so much, Eric. Thank you, Scott. What a treat. How can a community best support you? We got amazing feedback, and I can’t wait to share with you.

Scott Brinker:
Well, you can reach me at chiefmartech.com. That’s Chief Martech without the H at the end. That’s a fun branding choice.

Erik Charles:
And you can find me, one, on the HSC website, where you can track me down on their listing, but you can also just go to EricCharles.com.

Julia Nimchinski:
Awesome, thanks again, and now we move into the evolution of CRO.