SaaS Moats in the AI Era: Sell the Outcome, Not the Tool

0:00And then some dude with Claude says, "Hey, this is use my credentials to log in, screenshot that stuff and replicate it for my product." And they have it in 6 hours. [music] Somebody is going to go to your customer and sell the outcome without the employee having to do the clicking. And as far as I'm concerned, that might as well be you. If

0:18you don't do it, someone will. Guarantee it. [music] I make an AI receptionist and like the next day there's a gazillion AI receptionist. Like what Where is the defensibility in that offense? This is really the only way to sort of automatically stay ahead of this AI capability because it means someone can't just take AI, the frontier model, that's going to get better and better and better every year, plug it into the system, [music] um

0:42and then catch up with you instantly. So, one of the things that like you really should have taken away from Rob's talk is this notion that um that AI's everywhere and everyone gets AI. Um and sort of like what is what does that mean for um for you? And, you know, he went through a bunch of different things, like demand may go down or it could go [clears throat] up in some cases, but a lot of cases it could be going down. Supply is definitely going to go up for most of

1:12you. Um and and and that's going to be a challenge, right? And you might be sitting here thinking to yourself, "Oh, um like like maybe it's slightly nerve-racking if you never really thought about moats before, not really, uh certainly not very deeply. And perhaps you thought to yourself, well, [ __ ] the only moat that I really have is that it took a long time to build

1:34this software. And now that's not true anymore, right? Um but like, you know, may you live in interesting times is kind of a double-edged sword, but broadly speaking, like I want you to sort of take in take on board the fact that really like disruption and risk is something that you should want as like an upstart because in very stable systems, typically most of the gains end up staying with the incumbents because they have the most resources. It's only really like risk and disruption that gives opportunities to upstarts like us, particularly ones that aren't like super well-funded like like most of the people

2:12in this room. And so the way that I see like what are the offensive options in terms of like okay, here's AI, I have a SAS, what do I do? And I think like the first option is kind of like the obvious one, you just have to move faster. That's that's like, you know, and that could be enough in itself, particularly like if you looked at all these moats you're like, "Huh, I

2:33have six of 10. I'm amazing." Um and that's sort of like table stakes, I think. I think everyone needs to do that. I think that the sort of number that I'm hearing is that most mature SAS businesses are probably shipping features somewhere in the region of about at least three times as fast as they did prior to AI coming on [music]

2:49the table. The second part is okay, charge the agents. I think for for some of you here, and again, like Rob was saying, I don't really have a good sense of how many this is a good solution for, but certainly for some of you, it's going to kind of be obvious that people are going to want to use your tool with agents, and you're going to find out a way basically how to charge those agents, how to make it easy for them to get access. And I think also there's almost like a pair there was a couple of you know, SEO AEO talks. There's almost like a pair two parallel tracks for SEO,

3:20right? There's like how do humans find things and how do agents find things? Um and they may not be the same thing. So those are those are sort of the two things that I think about. But really, I think for most of you, the biggest opportunity here is to move up into labor markets. Um and and what do I mean by that? And and what am I what am I

3:40really saying? And the reason why I think it's the biggest thing is really that labor budgets are significantly larger than software budgets for most businesses. Like this is true. Like most pe- businesses spend a lot more money on salaries and people than they do on software. And like moving [clears throat] into labor budgets, like it might feel slightly bad because like aren't people losing their jobs if I do that? Well, that it was ever thus, right? Like the classic thing when MicroConf got started was like you'd be on the in the hallway and be like, "Yo, you know, what does your

4:11customer use if they don't use you?" We'd be like, "Oh, they use like spreadsheets and or just pencil and paper." But, you know, now they can get done in, you know, half a day what took two people 3 days. Well, this is the same kind of thing. It's sort of the supercharged version of it. And the the fact of the matter is a lot of SaaS is basically um here's a tool that you sell to a business owner and then that business owner gives it to a one of their employees who they're obviously paying a salary and then they click on the tool to get some outcome. And if that's your

4:41SaaS, I'm here to tell you right now that somebody is going to go to your customer and sell the outcome without the employee having to do the clicking. And as far as I'm concerned, that might as well be you. Because from my perspective, it's far from obvious to me that you're in a better position if you are some fresh-faced like Silicon Valley startup that has a $10 million token budget but zero customer experience, zero customer relationships, and all that's all they have versus what you have, which is an established SaaS, established customer relationships, you've been in the industry, you understand that they know you, that sort of thing. You're in a

5:21much better position to sell into this market than anyone who starts this from scratch. And like I'm here to tell you if you don't do it, someone will. Guarantee it. If you're getting value out of this talk, you should subscribe to the MicroConf YouTube channel. We're posting all the talks from MicroConf Europe. So, even if you couldn't make it to Iceland, you can still get a lot of the learnings from the event. We also have a back catalog of some amazing talks from our previous events. All

5:45right, let's dive back in. That being said though, like just moving into labor markets in and of itself, like, okay, well, I make an AI receptionist and like the next day there's a gazillion AI receptionist. Like, what Where is the defensibility in that offense, right? And And fundamentally, the way that I think about it sort of like this. You can imagine like this is where my sort of academic stuff comes in. This is when Rob starts going

6:10like, oh, here we go. There's like a you know, there's like a Imagine there's like a space of capabilities and this is down on the left here is what the AI can currently do. And it And I it says current LLM capabilities, but really what I mean is like either just Claude itself or like a startup with Claude to do certain things. And like right now they may be fine and then the red blob here is the capabilities that your SaaS currently has, right? And you're the the sort of angst in the room is like, well, how

6:36quickly does this get encompassed? Right? How quickly does does the does the LLM capabilities extend to the point where it encompasses all my SaaS capabilities? And how do I defend against that, right? That is the main problem. But, to degree, this was always always kind of the problem. Like, you know, it wasn't AI before, but it was basically like you couldn't just do nothing, right? You you always had to improve because otherwise your customers would come your competitors would come along and they would, you know, basically have better features than you and then everyone would go to that customer. So, this is not super new. So, to degree,

7:09sort of the old-school like quote-unquote learning loop is sort of the product loop, right? This is this notion that you're in space, you see what your customers are currently doing, you figure out, you know, what is it they need or maybe you talk to prospects and they're complaining, I'll buy if you build this thing. And based on that, you you use your judgment and you decide, okay, I'm going to build that thing and then you obviously you write the code and then you ship it everybody and then everybody benefits. And you expand the capability of the SaaS, right? And

7:37round and round it goes. As [clears throat] Colin actually said, as you can see, 42 minutes before I screenshotted this talk. Um basically, the loop is identify customer need, ship something useful, get feedback, identify customer need, ship something useful, get feedback, over and over and over again, right? Um and like the interesting thing about AI, other than the fact that I now get to put German collection on my one of my slides, is like AI makes the write the code part much faster, and everything

8:07else sort of stays the same. The problem though is everyone gets AI, right? And so there's no necessarily any defensibility in this. Like it AI because code is much faster to write now, and everyone gets it, every competitor, every in-house team, whatever, all get the same advantage. And the problem is, and this is where I was saying when when Rob was saying that he doesn't think like that like having a strong opinion necessarily is a moat, because there's like a you know, it might take you three or four months to really come up with the best possible solution based on deep knowledge about your your industry, your

8:43customers, your product, all that stuff, and then you build it very carefully, you craft it and ship it, and then some dude with cloud says, "Hey, this is use my credentials to log in, screenshot that stuff, and replicate it for my product." And they have it in 6 hours. So there's no automatic uh moat in just building features. So if you but but it's still stable stakes, so you still kind of have to do it. And like what I'm trying to encourage you to do is if you're doing that, now that you sort of understand, at least hopefully, what the moats are, use some of that additional

9:14uh code shipping ability to go for like steer towards the moats. Like understand what those moats are in your space, and which ones are still sort of going to to stay around, and then go for that. However, I think also you need to realize that even that may not be enough in and of itself. And I think but this is sort of where the learning loop comes in. And I think like the this relates to like this is why I'm saying you should go into the to the labor market because what you can do with once you go into the labor market is that basically the fact that your customers

9:50are using your product can make the product itself better ideally across customers. And so what does that mean look like? Well, it basically like the product does something, it responds to a customer ticket, it you know, closes an invoice, whatever it is that it does. It responds in some way, i.e. the customer pays the invoice or you know, is happy with the response or does whatever, you or the system is able to observe that success or failure of that outcome and then the next action improves and ideally automatically. So there's sort of a loop where basically the product itself learns over time and gets better and

10:28better and better. And then this doesn't have to be like like some some some products are like all of this. Like they're like, "Okay, every single thing here is uh the whole product is like this." For for a lot of you, there's like some parts of it where this makes sense and some parts of where

10:44it doesn't. And so the ideal learning loop here is the key thing though is that it doesn't leak to the outside world. The learning that improves the outcome for your customer stays with you. I it's not something that you're pushing to an external party. It's not something that can very easily export and get every value for themselves. It's also something that requires like a real world clock, i.e. observations that happen in the real world. It's not something that you can easily get by just simulating the outcome and catch up to you. I it's [music] not fast-forwardable. And then the ideal one, and that's this not everybody, the

11:18ideal one is that it improves across customers. So the fact is the fact that you've had a thousand customers for five years means that an upstart needs to basically live through those experiences in order to to catch up with you. So that's an example of this, and I see the clock is blinking, but for the

11:36record, I have 5 more minutes. Um imagine you had a collection SaaS, this is what the evolution might look like, right? You start off on the left here, and it's just like a classic dashboard. It just keeps statuses of like who paid what invoice, what offer did we make, when was this done? And then you think to yourself, oh,

11:52here's AI, let's sprinkle in some AI. The second part is then you add an AI assistant. So, it helps you figure out like let's draft a reminder, let's you know, the person still looks at it, but you know, like it helps you draft the email before you send it. Step three, and this is where you really start moving into the to the uh into the labor budget, is one where there's no human in the loop necessarily for this. Like the the person basically chooses who to contact, what offers to make in order to get the invoice paid, and maybe it does that based on just some hard-coded best practices that the

12:24business has, and then finally, the compounding AI worker is one where it basically It not only does that, but improves itself over time, so that it keeps improving. And really, from my perspective, this is really the only way to sort of automatically stay ahead of this AI capability, because it means someone can't just take AI, the frontier model, that's going to get better and better and better every year, plug it into the system, um

12:50and then catch up with you instantly. So, five tests, you know, does the product take the action? It has to be able to do that. It can't just be recommending. Uh can it see what happened afterwards? Does the outcome change what it does next? Can the lesson improve results for other customers? This is the ideal one, not everything is like that. And does the accumulated learning stay with you? Um

13:12and that is the crucial part. Um and talking about my why, this is a redfish that I caught in Louisiana. Um and then some resources stuff. [applause] Thanks for watching. There was a lot in that talk, so we turned it into a free field guide. The 10 attacks, the eight motes, and a scorecard you can fill out for your own business. Head to microconf.com/motes and we'll send you the guide. You'll also get the recording of the Q&A from after the talk. We're not posting that one on YouTube, so this is the only place to get it. And if you want to be in the room next time, in 2027,

13:46MicroConf is headed to Austin, Texas in April and to Dubrovnik, Croatia in October. Watching these talks on YouTube is great, but the real magic happens in person. You're in a room with about 250 founders who get exactly what you're building, and some of the best conversations happen in the hallway between the talks. The last seven, maybe eight MicroConfs has sold out, so grab your tickets now at microconf.com/events if you want to go. These events will

14:10sell out, and I hope to see you there. And if you already have your ticket, or you're just ready for another great talk, check out this one right here. I'll see you over there.