DEV Community

Loic Moncany
Loic Moncany

Posted on

Why I'm Not Scared of AI Killing My SaaS (And What You Should Actually Fear)

Why I'm Not Scared of AI Killing My SaaS (And What You Should Actually Fear)

Everyone's panicking. "AI agents will kill SaaS." VCs are writing essays about it. Twitter is full of takes. And yes — some of them are right.

But most of the fear is aimed at the wrong target.

I've seen this movie before

15 years building software. I watched the "SaaS killer" narrative run through at least five cycles:

  • Low-code was going to kill dev tools
  • No-code was going to replace everything
  • Web3 was going to decentralize all platforms
  • ChatGPT was going to replace Google
  • Now AI agents are going to replace SaaS

Each wave killed some things. Low-code genuinely ate into simple internal tooling. No-code took a chunk of landing page builders. The pattern is consistent: generic dies, specific survives.

What's actually different this time

AI agents aren't just hype — they're a real shift. But not in the way most people frame it.

The real change is a new question buyers ask before signing up:

"Can I just vibe-code this in a weekend instead of paying $99/month?"

For a lot of tools, the honest answer is yes. If your SaaS does one generic thing with no real lock-in, no proprietary data layer, no workflow depth — a decent developer can spin up an agent that covers 80% of it in a day.

Those tools are in trouble. Not because AI is magic. Because they were always just thin wrappers around APIs with a nice UI slapped on top.

What doesn't get replaced

I build EST8 (a real estate CRM) and ListingVid (AI video generation for real estate agents).

Neither of them is getting vibe-coded away. Here's why:

1. Workflow depth. EST8 isn't just a database with a UI. It's opinionated about how real estate agents manage deals — stage logic, reminders, client communication, document flow. An agent prompt can't replicate 6 months of iteration on actual agent workflows.

2. Vertical data. The value compounds over time. Historical deal data, client relationships, market patterns. An agent starting from scratch doesn't have that.

3. Integration context. Connecting to MLS feeds, e-signature tools, CRMs, and calendar systems isn't a weekend project. The integration surface IS the moat.

4. Trust. Real estate agents are moving serious money. They're not running their pipeline on a vibe-coded agent. They want something that looks professional, has support, and won't disappear.

The SaaS killers always target the generic

Every "SaaS is dead" wave hits the same category: horizontal tools that do one thing anyone could build. Expense trackers. Link shorteners. Simple form builders. Basic analytics dashboards.

If that's your SaaS — yeah, you should be scared.

But if you've gone deep in a vertical? Built something with workflow logic, proprietary data, real integrations? You're fine.

What you should actually fear

Not AI agents. These:

  • Staying shallow. If your moat is "we have a nice UI," that moat is gone.
  • Ignoring the shift. The tools that survive will use AI, not compete against it.
  • Wrong customer. Building for developers who can vibe-code? Hard. Building for real estate agents, restaurant owners, HR managers who can't? Much safer.

The move

Build something specific enough that a prompt can't replicate it in a day.

Go deep in one vertical. Make it opinionated. Embed yourself in real workflows. Use AI as a feature, not a threat.

The middle is dying. The edges are fine.


Building EST8 and ListingVid. Writing about SaaS, AI, and what it actually takes to build products that last. Follow if that's your thing.

Top comments (0)