Saw a Gong rep post on X about a new feature where AI listens to calls, extracts data, and populates CRM fields automatically.
My first reaction was wait, that's new? (sorry Gong 🤣)
I built that months ago with Claude Code. Updates CRM fields, progresses pipeline stages, pulls intelligence out of calls and feeds it into Salesforce. It wasn't some massive engineering project. I just built something I needed.
And apparently I wasn't the only one. Someone replied to the Gong rep saying they'd built the same thing in Claude two months ago.
The realization I had about a lot of these SaaS products is that you're a GitHub repo and a Railway deployment away from saving yourself $10K a year. You only use one feature on the product, and your manager before you bought it because they raised a round and didn't give a shit at the time.
The Gong rep replied with the enterprise objections: "Now do it for 80+ languages, multi CRM, and all security governance across 5,000 customers." And I found the exchange kind of funny because he's right to point that out. Enterprise readiness is tough.
But he's also kind of missing the point.
The Market Is Starting to Have the Same Gut Feeling
I've always had this instinct that SaaS is dying, at least from a publicly traded, PE-backed perspective. I've been hesitant to stake my claim there because I don't think it's dying in the short term. But I think the market is starting to have the same instinct.
Median year-over-year revenue growth across 58 public SaaS companies has dropped from 36% in Q1 2022 to 15% today. That's a 58% decline, falling every single quarter for three years straight. Maybe SaaS was overvalued to begin with but I think it's deeper than that. The iShares Tech-Software ETF has dropped 28% from its September highs. People are calling it the "Claude Crash." The name is dramatic, but the instinct behind it isn't wrong. The decline has a dozen causes. What matters is that institutional investors are starting to wake up to what I already know.
And it makes sense when you think about what most SaaS actually is. Strip away the branding and the sales deck and what are you looking at? An SQL table, some automations, maybe some proprietary ML infrastructure. A lot of that is hard to build, I'm not dismissing it. But I'm literally a half-braindead SDR who understands some basic concepts about computer science, and it's possible today for me to start replicating these features with raw code through tools like Claude Code, Cursor, and Replit.
The hard part is deployment, security, identity management, all the stuff that makes software "enterprise-ready." The moat question for public SaaS companies shifts from "can competitors build this?" to "can our own customers build this?"
The Barriers Are Real, But The Scales Are Starting to Tip
When you vibe-code something for your 10-person sales team, you're not thinking about SOC 2, HIPAA, role-based access controls, or what happens when your workflow breaks at 2 AM. These are the reason enterprise software costs what it costs.
But every one of those barriers is increasingly a standalone product you can buy. WorkOS gives you enterprise SSO through a few API calls. Vanta "automates" (yes I know this is a whole post on its own) SOC 2 and HIPAA compliance. Stripe handles PCI and billing. Supabase gives you row-level security for tenant isolation. Vercel and Railway handle deployment and scaling. You don't solve those problems anymore. You plug them in.
What I'm saying is not that these problems are solved. I'm saying they're solvable, and every quarter the cost of bolting enterprise-readiness onto something you built yourself gets cheaper. Not just in SMB. In mid-market and eventually in enterprise.
The Economic Problem Is Structural
The economic issue isn't really about features. It's about subsidies. Whether the balance sheets spell it out this way or not, if you've worked in SaaS you know how the game works: enterprise contracts effectively subsidize SMB and mid-market customers. The Fortune 500 accounts paying six and seven figures per year fund the R&D, the compliance certs, the support teams. The $5K/year SMB accounts rarely cover their own cost of acquisition and support. VC math makes it work anyway: spend three dollars to acquire a customer who generates one dollar today, bet they grow into five. The long tail funds the growth engine.
If SMBs start peeling off because practitioners can build "good enough" for their specific workflows, the subsidy model breaks. Enterprise customers won't pay more to compensate. They'll negotiate harder. The revenue doesn't disappear overnight, but the growth narrative does. 36% median growth down to 15%, every quarter, for three years.
What Happens Now?
I'm 33. The kids who are 10, 15 years younger than me have access to every tool I have. They're building with Claude Code and Cursor right now. When they're in my position and someone hands them a budget, do you think their first instinct is going to be "let me buy Gong at $30K/year"? Or is it going to be "let me build what I need now and buy the enterprise solution when we're actually at that scale"? Because the second one is already how I think about software, and I grew up before these tools existed.
If I'm an investor thinking ten years out, I'm not betting on enterprise SaaS. I'm watching the infrastructure layer underneath it. Railway and Vercel are already moving in this direction. Someone is going to package authentication, compliance, multi-tenancy, billing, and monitoring into a single platform and market it directly to this new generation of builders.
You built something great, we make it enterprise-ready, two grand a month. If I were running a Y Combinator batch and someone walked in and said "we're building the platform for SMBs and mid-market companies to self-host their own vibe-coded apps with security, identity management, and compliance baked in," I'd tell them to sit down and start talking. That market grows every time someone opens Claude and thinks, "I could just build this myself."
SaaS. Dying, molting, changing… not dead. Yet, anyway.