SaaS is not dead. The middle is.
There’s a lot of noise right now saying “SaaS is dead. AI agents killed it.”
That sounds dramatic. It’s also misleading.
What’s really happening is simpler.
Software is getting easier to build.
But making it work inside a real company is still hard.
That changes where the value sits.
The three layers of modern software
I like to think about today’s software stack in three layers:
Top layer – AI agents
This is the part that does the work.
Agents read emails, update records, run reports, write code, move data.
Middle layer – SaaS UI
Dashboards. Forms. Buttons. Workflows.
This is what humans click.
Bottom layer – systems of record
Databases. ERPs. CRMs. Finance systems.
This is where the real data lives.
Right now, value is moving up and down.
The top layer is getting stronger because agents can execute work directly.
The bottom layer is getting stronger because agents need clean, trusted data.
The middle layer — pure UI SaaS — is getting squeezed.
Why the middle is under pressure
For the last 15 years, the SaaS playbook was simple:
- Find a workflow
- Build a clean UI
- Add integrations
- Charge per user per month
This worked because software was hard to build and hard to change.
AI broke that assumption.
If an agent can log into your tools and complete a workflow end-to-end, the UI becomes optional.
If headcount goes down, per-seat pricing breaks.
This doesn’t mean companies stop buying software.
It means they stop paying for empty seats and pretty screens.
“Agents killed SaaS” — not really
Agents didn’t kill software.
They killed the idea that organizing a workflow was the moat.
Many products were really just:
- A thin UI
- Sitting on top of someone else’s data
- With switching costs that looked stronger than they were
When a plugin or agent can replace you in a weekend, the product was never deep.
That’s the uncomfortable truth.
Where the money is actually going
The money isn’t disappearing. It’s moving.
Here’s where it’s landing:
1. Agent platforms
Not per-seat.
Usage-based. Outcome-based.
Pay for what gets done, not who logs in.
2. Systems of record
Agents don’t replace databases and ERPs.
They depend on them.
Bad data in → bad actions out.
Clean data becomes more valuable than ever.
3. Security and governance
When agents act at scale, mistakes scale too.
Access control, audit logs, policy checks — boring, but critical.
4. Services
This part is underestimated.
When software gets cheap to create, companies don’t build less.
They try more.
Customization, migration, workflow design, integration — all still human-heavy work.
Creation is easy.
Implementation is still hard.
So what about Oracle, SAP, and the “bottom layer”?
This is where your question really matters.
On paper, companies like Oracle and SAP sit firmly in the bottom layer:
- Databases
- ERPs
- Financial systems
- Core business data
That should be a good place to be.
But markets don’t price theory.
They price fear and timing.
Right now, investors are nervous because:
- Pricing models are changing
- UI-heavy products look fragile
- AI narratives are moving faster than revenue models
That’s why you see pressure across enterprise software, not just one company.
It doesn’t mean the bottom layer is weak.
It means the market hasn’t fully priced how important it becomes in an agent-driven world.
The quiet winners
Some companies won’t look exciting on Twitter.
They won’t demo well.
But they will:
- Own trusted data
- Run critical systems
- Enable agents safely
Those companies age well.
Agents come and go.
Data systems last decades.
The real shift
This isn’t the death of SaaS.
It’s the end of easy SaaS:
- Thin UI
- Rented data
- Per-seat pricing
- Weak lock-in
What replaces it is harder, messier, and more real:
- Execution over display
- Data over dashboards
- Outcomes over seats
- Services alongside software
That’s not bad news.
It’s actually good news for people who’ve spent years working in real systems, real data, and real complexity.
Final thought
AI didn’t kill software.
It exposed which software was shallow.
The middle will shrink.
The top and bottom will grow.
And the hardest work — making all of this actually run inside real businesses — will matter more than ever.
That’s where the next decade is headed.