393: AI is a Threat to SaaS Multiples

I use it in a lot of my properties, and they're taking care of all the things related to money so that founders like me and you can focus on building the things that only we can build. And this is also a main topic of today's episode. But I digress. Paddle handles all the rest, sales tax, credit cards fitting, all of that. I highly recommend it, so please check it out at paddle.com.

Today I want to share something that's been on my mind ever since a conversation that I had with an investor friend of mine, Tyler Tringis. We were doing some fear setting, looking at how my business is currently performing, what it would look like if things went better, what it would look like if things went worse, you know, the kind of stuff, the conversations with your investor and mentor that make you really think about where you stand in the market and what's happening. And while I can't share all of the personal details of this conversation, there is one part that really stuck with me, and that's something that I think every SaaS founder needs to understand about what's happening right now in our industry or the many industries in which we build software as a service solutions. So for years, software as a service, SaaS, has been the golden goose of entrepreneurship. Right?

Founders have been building towards these beautiful acquisition multiples, six x, 10 x, 20 x to annual revenue in whatever shape or form that may be measured. There were a lot of x's and a lot of numbers. And these numbers weren't just wishful thinking. They were based on real market dynamics. That was actually the value of their companies.

See, there used to be a fairly sizable moat around SaaS businesses, which is why they got acquired for these prices. It wasn't just about understanding the market that you serve or having industry expertise, but building the actual technology of a software as a service business was a technical feat. And I don't just mean the product here. I mean everything that enables the business, like the sales pipelines, the marketing automations, all these little things that you had to clumsily and carefully assemble over many, many years. It took time to get things right, to even come up with things that nobody else may have ever built in that particular shape or form.

And that expertise was encoded in both the business model with recurring revenue, which is amazing and focuses on retention more than acquisition, and in the technical challenge of building the product and all these little things around it itself. So even when I started Podscan, and that is a year and a half ago, AI did not exist in a way that it does now. I still had to build all of this myself. Agentic coding was not a thing, at least not for me, and I consider myself almost cutting edge when it comes to that kind of stuff. It just wasn't good enough for what I needed, and I didn't go too deep with it.

There were already LLM systems that could kind of code for me. But in late twenty twenty three, when the first prototype happened and early twenty twenty four, all of that was still on the horizon. It wasn't there yet. So let me maybe share my experience building PodScan because it illustrates a shift here perfectly. When I started, I was able to spin up a functioning money making business within a day or two, at least on the technical side of things, using the Laravel framework.

And that was already so much faster than maybe ten years ago where I would have to kinda, again, assemble all of these components myself, build it myself, build an authentication, build a payment portal, integrate an API, something like this. Laravel is different. It is a framework that sits on top of the PHP language and has a very, very amazing ecosystem of plug ins and tools and services around it that really facilitate building a business on top of it. This is not an ad for Laravel, even though I highly recommend everybody who wants to build a business quickly to look into frameworks just like Laravel or Ruby on Rails, like these kinda established technologies that are focused on making things easy to build for people who wanna build businesses, not just software. But, again, it's less about the specific of the framework and more about the fact that there were frameworks there already in the first place.

And I had to pay for some licenses for some kind of more complex components, but most of it was free and open source. That's the interesting angle of the Laravel framework. And if you tune in over the next couple weeks, we may or may not have the person who built that framework on this podcast. So, yeah, tune in next week maybe. I use a lot of services, again, provided by the people behind Laravel to help me deploy it more reliably.

And for a couple hundred bucks initially and a couple hours of work, I would call myself a one x developer at best, maybe a 0.8, someone who can code reasonably well but not a wizard, I still after an hour or two, I had a running business just with all the basic components in place. Then figuring out how to build the actual product around it, that may take another day, a couple more weeks, but there was already a lot available to me that allowed me to hit the ground running. And it looked like a default template. It did the job, though. I started indexing a couple podcasts.

I made them searchable. I built the first downloading system, the first, like, transcription system. Within a week or two, everything was already ready for me to just implement this. And the arduous task of building infrastructure to capture more and more of the podcast ecosystem until I reached a % coverage, well, that took a while, a couple months. But initially, that first prototype was super easy to build and was all pre AI using default components.

And now, right now, and I've done this over the last couple of weeks, I could go to Bolt or Lovable or Cursor or Replit or whatever agentic system I prefer. I could say in a prompt, almost verbatim, build me a Laravel application using Laravel 12, implement an authentication system. I need to be able to log in with Google. I want a Paddle account attached to this, connected to an SQLite database, and then to the prompt, would add what I actually want it to do. I could tell the agent exactly this, and it would, in its agentic mode, likely give me a functioning system within five minutes, ten minutes max.

There might be a couple of changes I need to make, but the AI coding systems now are so good that they can't really be compared to what AI is if you only use it in a chat mode or even the early versions of Cursor or, what was it, GitHub Copilot. The systems that I'm using right now, which currently for me is inside of PHP Storms, which is IntelliJ's PHP IDE, I use Junie, which is their agentic coder. They call it delegate your tasks, focus on the results, which is exactly what agentic coding is. I just prompt it, and it uses the context of the the whole application to build the features that I want. Using these tools now includes things like self linting and self correcting.

Like, they check the code they wrote back into itself, then they verify that it is actually accurate and not hallucinated. And those hallucinations that it may have, they still happen. You just don't see it because the checking prompt built into the agent catches them and finds a non hallucinated version that actually works. Like, there's a lot of complexity behind the scenes that these agents just kinda go through, and it is very surprising how high quality the output of these tools can be. I can now spin up what used to take me a day or two, maybe a week, within a couple minutes.

It might take me an hour to customize the logo and whatnot, but the first initial version of a SaaS product is out in an hour and maybe a day if I want to spend some more attention on certain details. What used to take weeks then took a day, and it now takes an hour or just a couple minutes. And here's where things get interesting for us SaaS founders, and it gets a little scary for us too. If your moat was supposed to be your skill at building a product, implementing all the features, understanding business logic enough to build the technology, then obviously somebody else who has similar insight into your industry, understands the challenges and problems of your potential customers, and who may have the customer perspective and understand what jobs they want to get done, if they can prompt AI accurately today, they'll be at least 95% of the way to what a team with a CEO, a capable designer, a product owner, and an industry expert would have built five years ago. By today they also use these tools so they are ahead a little bit but the margin is slimming here.

You can now do all of this by facilitating AI systems. It's very easy and it's just good enough for prototypes and even full applications beyond. The moat is disappearing, and I think this has lasting repercussions for the investment and acquisition angle for Software as a Service businesses. Here's the question that is being asked in boardrooms all over the place right now: Why would we want to acquire a company for 10 times the revenue that they produce if we can set aside, let's say, one times that revenue that that company produces, give the project of building something just similar enough to a technical manager or a senior developer and pay for a fleet of AI agents to build the same thing. Could we get 80% there with 20% of the cost?

And I think we're getting there. I'm obviously not saying that this is going to work every single time. Like a software business has way more than just the product and the code base that is attached to it. Right? It needs a sales process and marketing process, customer support process, all that.

But if you already have these processes in house because you have a successful business that is looking to acquire or you just have something similar enough that it would warrant acquiring another product just like it, be it a financial acquisition if you want to keep it running or a strategic acquisition because you need the customers or the features of that product, you're not necessarily going to pay 10 x when you can try to build it yourself for a fraction of the cost first. You might still then acquire it, but maybe for less because you already spent some money. That also squeezes the potential multiple. And these conversations are actually happening in boardrooms right now, not just there, like even among the team members of these companies. They're like, well, this is a cool tool.

I think we should look into building exactly this for us. Let's spin up a couple agents and give them a well scoped prompt and see what happens and then take it from there. And this altogether puts negative pressure on the expected multiples for SaaS businesses. If the value of the technology is shrinking because it's so replicable, then the value of the company built around the technology shrinks as well, at least in the perception of the people who have to send you money to buy your company. So what can we do still to sell at high multiples, to build valuable businesses that are valuable enough to actually get acquired.

I think we need to do things that are not easily generated with AI, obviously, where AI just cannot do this for us, and that tends to be the human side of things. I think first, industry insight that cannot be trained is a massively important thing here. Like, we need highly connected salespeople. We need industry insights that is elusive to AI systems because it's insight that is so specific and earned from decades of being in an industry or working with certain people that is so encoded in your real experiences that AI could not have been trained on this. Or even if it has been trained on it a little bit, it hasn't been trained efficiently to synthesize this kind of information yet.

It doesn't have the instinct that a good salesperson has or that a CEO who really gets their customers has. AI can only generalize, but it might not be able to take a specific viewpoint to look into one particular customer perspective easily enough. There's immense value and unique human experience and the personal brand that will make something more valuable than just its technical implementation as well. If I build a tool and I build an audience around it, which I personally have, and I build a business on the overlap of my personal brand and the brand that tool represents, then all of a sudden the business and the product becomes much more interesting to acquire because it has a reputational advantage not just it's easily copied or somewhat easily copied technical implementation. That is very, very important.

If you build a product today and you're anonymous, you have a problem. I'll get to it in a minute. I think the second point that is equally important here is that we need strategic integrations and partnerships for our businesses as soon as we can. Existing collaborative integrations into other platforms, they will be much more important going forward. Like, there's value in having some kind of contractual relationship with another player in the field where they have plugged you as maybe one of their preferred vendors or you're listed on their blog as the tool to solve this particular problem best or they have an actual technical integration with your API.

Maybe you're featured in case studies that they have and they feature in case studies that you have. There is kind of this reputational alignment with other existing successful players in the industry, in the in the space that you operate in, or you share case studies, you show the relationships between your businesses and other businesses in the field, all of a sudden, this becomes invaluable because your product can be coned, but the many business dinners that you went to to find somebody to partner with and give and take negotiations to come up with a mutual agreement on how a partnership should look and the details of that, this cannot be done by an AI yet. There's always the caveat here. But at least it takes more time than an AI prompt can deliver results or meaningful value from. Right?

This is something that happens on an async between humans kind of basis, and that cannot be easily automated into an AI system. You could probably try to do a BizDev AI, but I'm not sure if the people that actually interact with that AI would be the people that a real business development person would try to build a relationship with. So having strategic partnerships and exactly mirroring those integrations between the businesses is very, very important just to stand out from those companies and those products who don't. And if you want to create a SaaS business that is protected from being easily cloned, well, you need to effectively do things that aren't easily automated because that's what cloning is. And this is all just relational stuff.

It's all interpersonal things that need to be deployed to protect your business from being a shallow AI copy. Personal brand works. I see this so much in the SaaS space. People who still sell for great multiples are people that are known. If you are the business that has an edge because there's a person attached to it, people tend to check out the human one first, the human business, the business with humans, and give it the benefit of the doubt over the anonymous one.

I highly recommend that software as a service founders be public as the founders of their business. Not only will this create a positive relationship with customers, like, obviously, if you are the person that runs the business and people who try out your products, you're interacting with them, that's amazing. Barely anybody does this. But it will also put you on the map for your founder peers and investors in that community that keep a watchful eye on people who go to conferences, who share interesting insights, who build interesting products, not just because they wanna acquire them, but because they have interesting human personalities behind them, and that makes them interesting to connect with, which then later down the road leads to better leads for an acquisition. And here's something else that is very interesting that more companies are missing at this point.

There's a lack of serendipity and opportunity when it comes to having human employees. I never thought I would have to use the phrase human employees in my life because, obviously, with AI agents, have virtual employees. And you see a lot of software development companies where people try a product because there's a colorful or helpful or kind personality working at that company, and they are just being relentlessly themselves on social media in good ways. Like, GitHub does this really well. They have a lot of people, obviously, who share a lot of their insights on social media.

And then just by being real people, I'm like, okay. Well, if those people built the software at GitHub, I will keep using that and not go to the alternative. Right? And a lot of other services do the exact same thing. Sentry comes to mind.

So equally interesting. It's just they have good employees. They have capable employees who have a point of view, and they share it. And for that reason, I kinda attach some kind of value and authenticity to the company. And it doesn't really matter what their personality is as long as people are showing up as their true selves on social media.

The businesses that they work for tends to get a lot of positive attention. And, you know, if you have people who have a point of view in a community, your business will benefit from this. And this falls flat completely if you don't have humans as employees. If all your company does is you as a founder giving commands to an AI system, which frankly is exactly what I'm doing right now with Podscan, but we're not at the size where employees would even be an option, then the opportunity for a developer that works for you to just say something cool on social media and people checking it out and checking out your product and suddenly they have a friend that might wanna use it, well, that serendipity potential is completely obliterated if your AI agent is the only one doing the work because they're completely virtual. They're not a real human personality.

They don't have a social media account yet, and they will never do anything but create code. Developers are people, and people are relationship builders and they are relationship amplifiers, right? They're proxies, they're nexus points for more and bigger relationships. If you replace all your devs or don't hire devs in the first place, then you're missing the serendipitous moment of them also working as this lighthouse to bring in attention from all around them, not just from other devs but from people in their vicinity that may or may not be in your industry or know somebody who is. So standing out and honestly just standing strong against the headwind of copycats and the permanent threat of being easily copied well that will I think define the next couple years of software as a servant businesses because multiples are dwindling because expectations and the reality of alternative solutions, like AI built clones and copies, I think we will see more competitive solutions being built instead of acquisitions.

And maybe acquisitions kind of as a second step after having attempted to build in an AI agentic fashion a competitive alternative. And there still are gonna be acquisitions, obviously. They still happen every day. But I believe that incumbents will actually be financially capable of using these extremely cheap and highly automated, but still well controlled and managed automation tools. Like, if you have a senior developer and you hand them a fleet of AI agents, if they know how to prompt and code review, they can do a lot of work or have a lot of work get done by just spending all of their time managing their fleet of agents.

And that is one person. It's not a team of 10. It's a team of one plus nine. It's a very different story. So they will use these tools to build things that are just good enough to be reasonably competitive with the company they would have acquired a couple years ago without a doubt, without even trying to build something.

Right? Five years ago, they would have just acquired the company for the multiple, and now they're like, should we maybe build this ourselves? We have all the tools. And I think this is obviously my personal opinion here, but I'm hearing this from my investor friends, from my founder friends. I'm seeing this in conversations around sales and acquisition multiples and prices and the disappointment of people where they thought the numbers that they wanted to get were easy to reach, but they're not.

They're much, much harder to get to. They require more of a sacrifice and, frankly, a more elaborate business to be sold for that much money at this point. It's a very interesting challenge for the next couple of years and I hope you take to heart mostly the idea of being present as a human being and focusing on allowing the human beings you work with to be an amplifier for the reputational value of your company because that's something that no AI agent can ever do for you, at least for the time being. I believe that there's a way out here in just to be a human being in a time of extreme automation. The artisan approach, just that we we are staying human approach, Remaining relatable feels like it's definitely the way to go for me.

And it's not gonna be the only way to go. It's not the only solution that helps here. Like, obviously, building a better business, like, out other things, better technology, use using AI like things as features or as amplifiers for the way you can crunch data and present new insights to your customers. All of this needs to be part of it too. And yet, I think we definitely need to think of being human as one of the things that we can do right now and all the time to protect ourselves and protect the value of the businesses that we create.

The future belongs to founders who can combine this efficiency gain from AI with the irreplaceable value of human connection, like wisdom about the industry and authentic relationships with people that you care about. The technical mode, I think, is disappearing. But the human mode, that is something that AI cannot replicate. So as you're building your SaaS business, ask yourself, what am I doing that only a human can do? What can I do today to improve the lives of my customer?

That is generally a good question to ask, but that is you. What can I do? Not what can AI do, but what can I as a human do? What relationships am I building that cannot be automated? What insights do I have that come from real experience, not some training data that was scraped from the Internet?

Those are the things that will protect your business and the value of it in this new world where anyone can build software, but not everyone can build the human connections that make a business truly valuable. And that's it for today. Thank you so much for listening to the Bootstrap Founder. You can find me on Twitter at avid kahl, a r v I d k a h l. And if you wanna support me in this show, please share PodScan.fm, my SaaS business, that may or may not be acquired one day, who knows, with your professional peers and those who you think will benefit from tracking mentions of their brands, their businesses, and names on podcasts out there.

PodScan is a near real time podcast database with a stellar API. We have 32,000,000 podcast episodes in the back now. The database is humongous. Please share the word with those who need to stay on top of the podcast ecosystem. Thank you so much for listening. Have a wonderful day, and bye bye.