We’ve now vibe coded 12+ apps into production at SaaStr. Some replaced things we were paying for. Some filled genuine white space where nothing existed. But two recent builds are different, because the implications go well beyond SaaStr.
We built our own AI VP of Marketing. And we built our own Customer Success and customer project management platform.
Here’s why — and here’s what it probably means for the industry.
The AI VP of Marketing: The Market Simply Doesn’t Have One Yet
We looked. Hard. We wanted an AI-powered marketing orchestration tool that could actually run marketing for a niche events and media business. Not generate copy. Not draft social posts. Actually run marketing — prioritize campaigns, allocate budget, manage sponsor pipelines, tell us where we’re behind on revenue and attendance goals, and what to do about it.
That product doesn’t exist in the market yet.
So we built it. We call it 10K internally. We gave it access to 10+ years of SaaStr data — what campaigns worked, what sponsor packages converted, what content drove ticket sales, what attendance patterns look like. We connected it to Salesforce via APIs and Zapier. Every morning, it tells us exactly where we stand against our goals, what we should do today, and where to adjust.
It even tells us when we’re behind, without being asked.
No off-the-shelf tool could have done this. Not because the tools are bad — but because they don’t have our data. And they’re built for horizontal marketing teams, not a niche B2B events business running on three humans and 20+ AI agents.
This was a genuine white space build. Nobody lost a deal here. The product simply didn’t exist to buy.
The Customer Success Platform: The Market Has Plenty of Options — All Wrong
The Customer Success build is more important for you to understand.
There are plenty of CS platforms out there. Gainsight. Totango. ChurnZero. Planhat. Dozens more. We evaluated them. They all share the same fundamental problem: they were built for a different era, and almost none of them have meaningfully integrated AI into their core workflows yet.
For our use case — managing sponsor relationships for a niche events business — they were also built around assumptions that don’t fit. They assume large CS teams. They assume standardized customer journeys across hundreds of accounts. They assume you want dashboards and health scores and playbooks designed for a 20-person CS org.
We have three humans. We run hyper-specific, highly customized relationships with event sponsors. What we needed was a CS and project management tool that understood our workflow, used AI to surface what needed attention, and moved at the speed we actually operate.
So we built it.
It’s not a general-purpose CS platform. It’s ours. It fits our niche exactly. And it does it with AI built in from the start, not bolted on as an afterthought.
Two Vendors Lost ~$20k Each. They’ll Never Know.
Here’s the part that matters for everyone building B2B software right now.
Somewhere out there, two vendors just lost approximately $20,000 a year each in ARR. Maybe more. We don’t know exactly who — we never got far enough in the process to land on a specific tool we’d have bought. But we know the category.
Those vendors will never know we existed. We’ll never show up in their churn data, because we never became a customer. We’ll never show up in their competitive analysis, because we didn’t leave for a competitor — we built something ourselves.
This is how AI is cutting into SaaS revenues. Not just from AI-native competitors taking share. Not just from consolidation. But from customers who simply stop buying the horizontal platform because they can build something that fits better, faster, and cheaper than ever before.
The Threshold Has Changed
The traditional build vs. buy calculus used to be simple: buy unless you have no other option. Building software was expensive, slow, and required engineering talent most operators didn’t have.
Vibe coding has broken that calculus.
The cost to build a functional internal tool or a niche-specific platform has dropped by an order of magnitude. You don’t need a team of engineers. You don’t need months of development time. You need a clear problem, access to an AI coding tool, and the willingness to maintain what you build.
The threshold used to be: only build if you absolutely can’t buy. Now it’s closer to: build if what you can buy doesn’t fit well enough.
For niche use cases, “doesn’t fit well enough” covers a lot of ground.
What This Means If You’re Building B2B Software
The threat isn’t just from competitors building better versions of your product. It’s from your customers deciding they’d rather spend 20 hours building something specific than $20,000 a year on something general.
That math is now real for a meaningful percentage of your customer base — not the majority, but not trivially small either.
A few things worth thinking through:
- The horizontal platform is under more pressure than you think. If your product is built for the average use case, every niche customer is now a potential churn risk — not to a competitor, but to a vibe-coded internal tool they’ll build themselves and never tell you about.
- AI integration is now table stakes, not a differentiator. Both tools we replaced (or would have bought) suffered from the same fundamental problem: they weren’t AI-native. If your product was built before 2023 and you haven’t deeply rebuilt around AI, you’re at risk. Not from the AI features that show up on your pricing page — from the fact that customers can now build AI-native tools from scratch faster than you can retrofit them.
- The customer who builds their own thing is invisible. They don’t show up in churn. They don’t show up in competitive intelligence. They don’t fill out your loss analysis surveys. The only signal you get is that a pipeline deal never closes — and you’ll probably attribute it to something else.
- Niche is where this hits first. We’re an extreme case — a small team with a very specific workflow in a narrow vertical. But “small team with specific workflow in a narrow vertical” describes a huge portion of the B2B market. Every category has its niche customers. And those customers now have real options.
We’re Not Stopping Buying Software
To be clear: we still buy plenty. Our CRM, our outbound SDR agents, our email platform, our analytics — all third-party tools. The 90/10 rule still holds. Buy what exists and works. Only build the 10% where nothing fits.
But that 10% is bigger than it used to be. And the definition of “nothing fits” now includes “the market has options but none of them are AI-native enough for how we work.”
The two vendors who lost our $20k each aren’t bad vendors. They built real products. The market just moved faster than their roadmaps, and we were able to build around it.
That’s the thing about AI eating software: it doesn’t always announce itself. Sometimes it just quietly removes you from a deal you never knew you were in.
