Is SaaS dead?

March 2026, in the middle of Ho Chi Minh City, Vietnam. It’s the afternoon and outside temperature is 33 degree celsius, so I’m staying inside my AirBnB to work on BlogSEO.

I had a successful relaunch on that platform a week ago, so I’m casually browsing There’s an AI For That and land in the SEO category which is where my product is listed. I see a new tool I had never heard of before. I click on it, check it out and start watching the founder’s video.

2 minutes in the demo video, I notice something strange. The dashboard looks oddly familiar.

I decide to test the tool to investigate a bit more. I land on the onboarding page.

The design was exactly the same as mine. Same font. Same colors. Same page structure. Same copy.

I was staring at a clone of my own product.

At first I was very angry. Months of building, designing & polishing and some random guy just copy pastes the whole thing?

But now I feel grateful because someone admired BlogSEO’s design enough to steal it. And god knows I genuinely care about my product’s design, so this is truly is a nice compliment.

Turns out, the founder isn't a developer. His SaaS was 100% vibe-coded with AI (a bit too fast as well), no engineer ever reviewed a single line.

The results: Bugs everywhere. Potential security breaches for each feature. Low quality output. Powerful vibe coding tools don’t make you good at product.

And yet, he still made some money from the tool (probably because people can’t cancel which is illegal but no one must have told him). Which raises the following question, assuming the models are going to keep improving:

Code has become commoditized.

Claude Code with Opus 4.7 and Codex with GPT-5.5 can take a vague problem description and ship a working solution end-to-end. Not perfect, but good enough that entire categories of software are starting to look exposed.

Why pay a $100/seat CRM subscription when it's just a SQL database with integrations? Why pay for a scheduling tool when a non-developer can vibe-code one over a weekend? The "build vs buy" math is shifting fast, and "buy" is losing ground every month.

So if anyone can build your product, what's left to compete on?

The rest of the value chain. The parts AI can't commoditize (yet).

Here's the uncomfortable truth I've learned the hard way: I've watched inferior products make 20x my MRR because their founders were better at marketing.

Same problem space. Worse UX. More money in the bank.

Building a great product is necessary for retention and LTV, but a great product nobody hears about is just an expensive hobby. And if you have excellent salespeople, structured upsells alone can compensate LTV for an average product.

Now here's the part most engineer-founders don't want to hear: structuring a sales team is much easier than building a truly great product. Great product design takes years of taste-building and an obsessive attention to detail that very few people I've met actually have. Sales playbooks, on the other hand, are well-documented and repeatable.

So if you're looking for the path of least resistance to a moat, it's not the product. It's distribution.

7 Powers: The Foundations of Business Strategy is a book by Hamilton Helmer that lists 7 different conditions that can create persistent differential returns, in other words, a durable competitive advantage, a.k.a “moat”.

Miss all 7 powers? You have no moat. You’re just waiting for your turn to get cloned.

Here’s the list:

The pain a customer would feel if they left you. The higher the pain, the stickier your revenue.

For SaaS, this usually shows up in three forms: data lock-in (years of analytics, custom dashboards, historical records), workflow lock-in (your tool is wired into 12 other tools via integrations), and learning lock-in (the team spent 3 months getting fluent in your UI).

SaaS example: Webflow. Spend months (or thousands of dollars) building your Webflow site, then try to leave. You can't export it. Stop paying, and your website disappears. When you realize this, it’s usually too late. Huge vendor lock-in.

Can you do it bootstrapped? Yes. This is the moat you can start building on day one. Even a small product can build switching costs through good integrations, data accumulation, and saved configurations.

A brand becomes a moat when customers pay more, or trust you faster, because of who you are, not what you do. Two products can do the exact same thing, and people will still pick the one with the stronger brand.

Brand isn't your logo. It's the gut feeling people get when they see your product. "These guys have taste." "These guys are serious." "These guys get it."

SaaS example: PostHog. Functionally, it's product analytics, same category as Mixpanel, Amplitude, and a dozen others. But PostHog's brand (open source, dev-first, hedgehog mascots, being weird) signals "this is built by engineers who actually use it" so clearly that technical teams default to PostHog without seriously evaluating the alternatives.

Can you do it bootstrapped? Yes, but it’s very slow. A brand is one of the hardest thing to build but it’s also one of the strongest moats because it gives you pricing power. Branding compounds over years of consistent voice, design, and customer experience. You can't sprint it, and you need to be intentional about it. , so even modest taste stands out fast in crowded categories.

You adopt a business model that incumbents can't copy without cannibalizing themselves. The bigger the incumbent, the more painful the copy.

SaaS example: Figma. Adobe's design tools were desktop-first and license-based, which is how they made billions. Figma launched browser-first and collaborative, treating design like Google Docs. Adobe couldn't follow without admitting their entire desktop suite was the wrong architecture and undercutting their own pricing. They eventually tried to buy Figma for $20B. The deal got blocked. Figma is now the default tool for design.

Can you do it bootstrapped? This is the underrated one for small players. Find a workflow where the big incumbent makes too much money the old way to attack you. Pricing models work too: usage-based pricing is hard to copy if your competitor's revenue depends on per-seat licenses. Marc Lou’s Zenvoice is another good example of counter-positioning.

Exclusive access to something competitors can't get. A patent. A key person. A dataset nobody else can build.

SaaS example: Google. Their cornered resource isn't the algorithm anymore (Bing and others have caught up technically). It's 25+ years of click data: which results people actually clicked, scrolled past, or bounced from across trillions of searches. Nobody else can replicate this dataset because nobody else had a search engine watching the entire web for two decades. This is why every "Google killer" loses, even when their tech is good.

Just look at how fast they caught up with OpenAI on the AI race once they decided to take it seriously: they had the data, the talent, and the distribution sitting there the whole time.

Can you do it bootstrapped? Mostly no. Unless you have proprietary data nobody else can scrape or replicate, this isn't your moat.

Your team executes in a way competitors can't replicate, even if they know exactly what you do. Toyota's manufacturing process is the canonical example. Knowing the playbook isn't the same as running it.

SaaS example: Honestly, rare in software. The clearest examples are deep-tech companies like TSMC, where decades of accumulated manufacturing know-how let them produce chips that Intel and Samsung still can't match despite spending tens of billions trying. In SaaS, this kind of moat barely exists. Most "process power" claims are really just engineering culture and good internal tooling, both of which AI is rapidly commoditizing.

Can you do it bootstrapped? No. This is a power that emerges over decades of organizational learning. Don't pretend you have it.

Your costs go down faster than competitors' as you grow. More customers, lower unit cost.

SaaS example: AWS. Every new customer makes the next server cheaper to operate because you can share the ressources. Smaller cloud providers can't match the pricing.

Can you do it bootstrapped? Not really at the start, but worth thinking about for your endgame. If your unit economics improve with scale, you can outlast competitors in a price war.

Your product becomes more valuable as more people use it. Classic two-sided marketplaces, social networks, communication tools.

SaaS example: BlogSEO. Our backlink exchange network becomes better and better as we get more websites on board (backlinks become more contextually relevant and higher quality because there is more websites to choose from).

Can you do it bootstrapped? Hard, but possible if you build network effects into the product from day one. Most SaaS don’t have natural network effects, so don't force it.

Here's a thought that's been going around lately.

If anyone can clone your UI in a weekend, and AI can rebuild your features in days, maybe the UI itself is becoming a liability.

The argument goes like this: half the time you use software today, you don't open the dashboard. You ask Claude or ChatGPT to do the thing for you. "Schedule a meeting with Marc." "Pull last month's revenue from Stripe." The agent calls the tool, returns the answer, you never touch the underlying app. Buttons, forms, modals, dropdowns: all of it is just a translation layer between humans and APIs, and LLMs don't need translation.

Follow this logic to its conclusion and the chat interface becomes the last UI you'll ever use. The implication for builders: stop polishing a frontend nobody will see. Ship a CLI. Ship an MCP server. Make your product accessible to agents first, humans second.

I partly agree, but I think the takeaway is more nuanced.

Agents will absolutely eat the routine stuff (fetching data, drafting messages, scheduling). But UIs aren't going anywhere for the work that actually matters. Three reasons:

So my bet: chat becomes a layer on top of UIs, not a replacement. Most people have spent 20+ years clicking buttons and dragging things around.

That muscle memory doesn't disappear in a year because OpenAI shipped a new feature. Behavior change is slow, even when the technology is fast.

The play isn't "abandon the frontend." It's build the API layer and the UI. Make your product accessible to both agents and humans. Pick one, lose to the team that did both.

That said, "agent-first" products are starting to emerge, and some of them are genuinely impressive:

I met Thomas in 2022 at my first job as a software engineer at Joko. We worked on the tech team together, and he's the person who taught me the most stuff I still use today. Back then, he knew everything about frontend and web dev. Now he's the one showing me the latest AI tools I didn't know I needed (Wispr Flow being the most recent. Seriously, if you spend 4+ hours a day typing, give it a try. It will change your life).

Along with his two co-founders, Thomas is building Wonda, the ultimate CLI for marketing. If you're an active Claude Code user, you can now run your entire marketing stack directly from it. Install the CLI, set the skill, you're good to go. I've been running UGC AI videos for my Meta Ads campaigns with it lately, and the results have been genuinely impressive.

If you're a solo founder drowning in marketing tasks while trying to ship product, Wonda is worth a serious look. It's exactly the kind of agent-first tool that ties back to everything I just wrote about: a great product that meets you where you already are (your terminal), without forcing you to learn yet another dashboard.

It's been a while since the last newsletter (4 months…), and a lot has happened.

Since the last edition, I've more than doubled BlogSEO's MRR. Customers keep coming in, retention is solid, and the product has matured a lot thanks to the feedback from early users (thank you to everyone who took the time to share theirs).

That said, I'm behind on the ambitious targets I set for myself at the start of the year: $1M ARR by the end of the year.

Which is exactly why I'm launching BlogSEO on Product Hunt today.

It's the biggest distribution lever I haven't pulled yet, and the timing finally feels right: the product is stable, the onboarding is tight, and the offer converts. Time to push.

Until next time, keep scaling! 🚀

Vince