Is AI About to Kill SaaS?
No, AI isn’t about to kill SaaS, but it is absolutely killing lazy, rigid SaaS. The ground under software feels shaky because customers suddenly know they have options: vibe-coded internal tools, AI agents, plug‑and‑play automations that used to require whole product teams.
Yet if you look at what’s happening with serious players, it’s less “funeral for SaaS” and more “evolution pressure is now at 11/10.” Databricks, for example, just crossed a $5.4 billion revenue run rate with 65% year‑over‑year growth and over $1.4 billion coming purely from AI products, which is not exactly the profile of a dying software company.
I’ve had enough conversations with operators to notice a pattern: nobody’s trying to rip out their core systems of record overnight, but everyone is suddenly questioning whether they should keep paying big invoices for tools that don’t adapt.
One founder told me they rebuilt a pricey internal productivity stack over a weekend using a couple of AI‑assisted coding tools and some off‑the‑shelf APIs; it wasn’t pretty, but it worked well enough that the renewal felt… awkward. That’s the new tension: “good enough and mine” versus “polished but inflexible and expensive.”
AI, Vibe Coding, and the New Buyer Mindset
AI isn’t killing SaaS; it’s changing how buyers think about what “software” even is. Vibe coding, getting something useful built fast with an AI assistant, a prompt window, and a vague idea, has become the gateway drug to custom software for people who used to be stuck in spreadsheet hell.
You see it in boardrooms and Slack threads:
-
“Couldn’t we just build this internal tool ourselves?”
-
“What if we wire this LLM to our data warehouse and skip the extra product?”
Once a finance lead watches someone spin up a custom dashboard in an afternoon, patience for a clunky SaaS UI drops to zero. I spoke with a sales leader who admitted they’d tried a whole zoo of AI‑powered internal tool builders, literally double digits, just because it felt faster than getting a vendor to tweak a workflow.
That’s the vibe right now: experimentation first, contracts later. Of course, what people don’t always see is the tax they’re paying in hidden fragility: brittle scripts, no proper logging, questionable auth, and dashboards that break when a field name changes.
But you don’t feel that on Day 1. On Day 1, it feels like magic, and you feel clever.
Systems of Record: The Real Moat
AI is more likely to eat SaaS interfaces than SaaS databases. The tools that store your contracts, HR records, customer histories, and billing data, those “systems of record”, are painful to rip out, and nobody sane is eager to migrate them just to try a new UI vibe.
That’s exactly why Databricks leans so hard into its data platform as the foundation and layers AI on top through things like Genie, a natural‑language interface for querying data. The lesson for SaaS teams is pretty clear:
-
If you’re just a thin workflow on top of someone else’s data, you’re vulnerable.
-
If you are where the data lives and is trusted, you can swap the UI, bolt on AI agents, and still be the backbone.
I’ve seen companies try to “vibe‑replace” their system of record and quickly realize they’ve walked into a multi‑month data migration swamp.
Most back out and instead ask: “How do we get a smarter, more flexible layer on top of what we already have?” That’s the opening for smart SaaS: become the stable core that everything AI‑ish plugs into.
Why AI UI Matters More Than Feature Checklists
AI isn’t just another feature; it’s a reset of how people want to interact with software. Databricks’ Genie is a good example: it lets non‑technical users ask questions in plain language instead of writing queries, which drives more usage of the underlying platform.
When you can talk to your tools instead of learning their quirks, the old moat of “specialized product experts” starts to disappear.
That’s the uncomfortable part for classic SaaS vendors. For years, having an army of Salesforce admins or SAP consultants was a moat: once people spent their careers mastering those UIs, they were locked in. In an AI‑first world, the product that wins might be:
-
The one with the best conversational interface.
-
The one AI agent can navigate easily through APIs and structured responses.
-
The one that hides complexity instead of training people to live inside it.
A friend who works on a large enterprise tool joked that their most passionate power users now “only touch the product to fix what the AI messed up.” Slight exaggeration, but it captures the shift: humans want to spend less time clicking around inside SaaS, not more.
From “Use Our Product” to “Build On Our Platform”
The companies that survive this shift will act less like sealed products and more like platforms with Lego bricks that customers can rearrange.
We’re already seeing that in how some vendors pitch themselves: “Bring your weird workflow, we’ll bend around it.” Instead of forcing teams to adapt to a rigid SaaS, they expose APIs, low‑code builders, and even full‑on “vibe coding” sandboxes on top of their system of record.
It’s not just a nice‑to‑have. It touches the metrics every SaaS leader cares about:
-
Higher usage because the tool actually matches local workflows.
-
Lower churn because ripping things out means throwing away custom logic, not just a license.
-
Easier expansion because the same platform can power new micro‑apps and internal tools.
One product leader told me their turning point was giving customer success teams the power to build small, targeted apps for frontline users on top of their own platform.
Suddenly, instead of begging users to adopt a giant monolithic UI, they delivered just what each role needed, and adoption quietly doubled. Nothing fancy, just respectful of how people actually work.
Is AI Going to Kill SaaS?
AI won’t kill SaaS; it will expose which SaaS businesses were coasting on inertia. The market has already started to price that in, application software names have lagged as investors worry about AI‑driven disruption and multiple compressions in the sector.
At the same time, companies that embrace AI as a new interface and a new kind of platform layer are posting numbers that look anything but doom‑y, as Databricks’ growth and AI revenue run rate make pretty obvious.
If you’re building or running a SaaS product, the real question isn’t “Will AI eat us?” It’s closer to:
-
Are we a real system of record or just a pretty front end?
-
Can users and AI agents build on us, or only around us?
-
Do we make people feel powerful with AI, or trapped in old UX patterns?
AI changes the default from “buy software and live with its limitations” to “shape software around how we actually work.” The SaaS companies that lean into that, who are willing to hand the fork to their users and say, “build on us,” are the ones that will still be standing when this hype cycle settles down.
Disclaimer:
This article discusses trends, opinions, and observations about artificial intelligence (AI) and the Software-as-a-Service (SaaS) industry. The views expressed are for informational and educational purposes only and should not be interpreted as financial, investment, or business strategy advice. Market conditions, company performance, and technology adoption can change rapidly, and readers should consult official company reports, industry research, or professional advisors before making business or investment decisions.
