A lot of people are discussing whether or not SaaS is in trouble. But in my opinion, and experience, the “end of SaaS” debate is missing the point entirely.
One camp says SaaS is dead because anyone can vibe-code any app now. The other camp fires back:
- “That’s naive.”
- “Real SaaS is hard.”
- “You need authentication, security, hosting, uptime, compliance, database management.”
- “You can’t just prompt your way to Salesforce.”
Both camps are wrong. Or rather, I think they’re arguing about the wrong thing entirely.
The real threat to SaaS isn’t that someone will copy your big, complicated and well established product. It’s death by a thousand cuts.
Let me give you an example.
I was paying for a SaaS, a fairly expensive one, that finds podcasts relevant to my expertise and pitches me as a guest. It scraped podcast directories, matched shows to my profile, drafted outreach emails, tracked responses. Polished interface, useful tool, every month I was debating whether it was worth the cost.
A few evenings ago, on the sofa, I decided to see what Claude Code could do.
Around midnight, I had a working system.
- It pulls podcast data from open-source directories instead of paying for expensive proprietary data.
- It stores everything in Notion, a tool I already use every day, instead of some purpose-built database I have to learn and maintain.
- It analyzes episode content against my background, writes personalized pitches that reference specific episodes, and drafts them straight into Gmail.
I’ve sent 48 personalized pitches through it. My cost per pitch is about five cents. The SaaS was charging me hundreds per month for roughly the same output.
Now, could I have built the SaaS itself? The whole multi-tenant platform with user accounts and billing and onboarding flows? Of course not. The complexity crowd is right about that.
But I didn’t need to build the SaaS. I needed to build something that gives me a comparable outcome. For a crowd of one, me, who makes the decision of how polished the system needs to be.
The outcome turned out to be a few API calls to free data sources, a prompt, and tools I was already paying for.
The SaaS-is-fine crowd keeps defending the moat of complexity, the infrastructure, the auth layer, the scale, as if that’s what customers are paying for.
But as a customer I am not paying for your infrastructure. I’m paying for an outcome. And when I can reach that outcome in an afternoon with tools I already have, the infrastructure becomes irrelevant.
No single person is going to out-build Salesforce. Nobody is vibe-coding a competitor to your SaaS with all its intricacies and edge cases. The complexity defenders are right about that, and yet they’re completely missing the point.
The danger is a thousand people, each solving their own specific version of the problem, in their own way, with their own tools. Not building a competitor, just cancelling their subscription because they don’t need it anymore.
A lot of SaaS products are, at their core, a workflow wrapped around an API call. Find data, filter it, transform it, present it, maybe send it somewhere. This kind of thing used to require a team of engineers and now requires a conversation with an AI coding agent and an afternoon of free time.
My podcast pitcher isn’t nearly as polished as the expensive SaaS version. There’s no progress bar, no settings page. It runs in my terminal. But I genuinely do not care, because it does the thing I was paying for, with data sources that are free and tools I already own.
I keep seeing this pattern. Not just in my own work, but in conversations with people running businesses, managing teams, solving problems. People who would never have called themselves technical, myself included, are building solutions to nagging problems that used to require a subscription to someone else’s software. Not because they learned to code but because the barrier dropped to nearly zero.
A year ago, I had a vague notion of what GitHub was useful for, but I also knew it wasn’t for me. I used Google Drive, iCloud and Notion. Now, my GitHub account is busier, and more frequently used, than any of the other tools mentioned. My day isn’t spent in email and browsers, it’s spent in Terminal and Obsidian.
The founders I’d be worried about aren’t the ones whose product is hard to replicate. They’re the ones whose product’s outcome is easy to replicate. And every month, more people are discovering they can reach that outcome themselves.
Not one big competitor. A thousand quiet cancellations.
This is the kind of shift I help organizations navigate as a fractional Chief AI Officer. If your industry is feeling this pressure, let’s talk.