Let me start with where I actually land on build-vs-buy:
I’d rather not build or vibe. Almost ever. If you can buy it — buy it.
The 90/10 rule. ~90% of the time, you should buy. Even more, really. If you can find it, and it meets 90% of your needs, buy it.
Software is cheap relative to the time it takes to build, and more importantly, maintain. Even as build costs have come down with AI, maintenance takes time. Vendors who specialize in a thing will out-execute your in-house maintenance of that thing. Maintenance is a tax that compounds quietly. Distraction from your actual product is real.
For 6+ years, that’s exactly why we paid ~$4,000/year for a fairly slick but simple newsletter building tool. It worked. We were happy with it. No real complaints. Buying was clearly the right call.
This week, we replaced it. In about 2 hours of vibe coding.
Not because we suddenly turned into builders. Not because we cared about the $4K. We didn’t, really.
We replaced it for one specific reason: the product had frozen in time, and building our own would save Amelia 90+ precious minutes a week.
That’s the whole story. And it’s a story every B2B SaaS founder should think hard about right now.
Why We Built. And Why It Wasn’t Remotely About the Money.
Three things flipped the calculus:
1. The vendor had shipped nothing new material in 5+ years.
No autopopulation from our content. No AI summaries. No API integrations to surface top tweets or top videos. The product in 2026 looked basically identical to the product in 2020.
2. We could build something that actually fit our workflow and added AI capabilities in 2 hours.
The new tool autopopulates the entire newsletter from our content. It uses AI to summarize posts (the posts themselves I still write, that part doesn’t get outsourced). It pulls top tweets via the X API and top SaaStr videos via the YouTube API. Automatically. Every send.
3. It saves Amelia our Chief AI Officer 90+ scarce minutes a week.
This is the part that actually mattered.
Amelia is SaaStr’s Chief AI Officer. Her time is not fungible with $4K. Her time is the single most valuable resource we have. Saving her 90 minutes a week is worth substantially more to SaaStr than the entire vendor relationship.
If the vendor had shipped these features over the last 5 years, none of this happens. We’d still be happy customers. Buying would still be the right call.
But they didn’t. So we vibed it. Actually, I did to save her the time.
The Lesson Isn’t “Vibe Coding Kills All SaaS”. But Point Solutions Frozen in Time Are Very Vulnerable
The easy, lazy take here is “AI makes everyone build everything themselves now.” That’s wrong, and it’s not what happened.
The actual lesson is narrower and more important:
Point solutions that don’t keep up can now be bypassed in the AI era.
Not all SaaS. Not the hard stuff. Not deep platforms with real data moats, real integrations, real workflow lock-in, real compliance complexity. Those are still firmly in 90/10 territory. Buy, almost always. We’re not about to vibe code a replacement for Salesforce or Snowflake. The 90/10 rule is alive and well there.
But for point solutions that have stopped evolving (single-purpose tools doing one job, often priced $200 to $2,000/month), the buy-vs-build math has shifted. Not flipped. Shifted.
In 2020, a non-technical operator was never going to build their own newsletter tool, no matter how stagnant the vendor. The cost of getting an engineer to build and maintain it was 10x the subscription. Buy won every single time.
In 2026, on a quiet weekend with Replit and Claude, building it yourself is a perfectly reasonable choice if (a) the vendor has frozen in time and (b) the workflow savings are real.
That’s a meaningful change. And vendors who haven’t noticed it are going to lose their best customers, quietly, at renewal, to in-house replacements.
Stealth Churn Is The Form This Takes
Here’s how this kind of churn actually shows up in your numbers:
- Customer doesn’t complain
- Customer doesn’t downgrade
- Customer doesn’t respond to CSM check-ins any differently than usual
- Customer quietly builds something on a weekend
- Customer cancels at renewal with a one-line email
You won’t see it coming because there’s no signal in the data. Usage stays flat right up until the cancel.
And the customer who built it themselves usually ends up with something better suited to their actual workflow than what they were paying for. Even if you ship a competitive feature 18 months later, you’re not winning them back. They’ve got something they like more, that fits them better, that they own.
What To Do If You Run a Point Solution
If you sell a point solution to technically capable customers, here’s what I’d actually do:
1. Be honest about which of your features could be replicated in an afternoon of vibe coding. Don’t be defensive. Just look. Then assume your top 20% most-technical customers are already at least thinking about it.
2. Ship. Much, much faster than in the past. I cannot say this enough. If your product hasn’t changed materially in 18 months, you are slowly losing your best customers. The “stable, reliable, no-surprises vendor” positioning that worked great in 2018 is now a liability.
3. Build defensibility around the things vibe coding can’t replicate fast. Proprietary data, deep integrations, multi-tenant complexity, compliance, network effects, workflow that crosses many internal systems. Lean into those things.
4. Bake AI into the core of the product, not as a feature flag. Customers who can vibe code will not pay you a premium for AI features they could build themselves in 2 hours. AI has to make your product structurally better in ways a single builder cannot replicate on a Saturday.
5. Watch your most sophisticated customers’s DAUs and WAUs and MAUs. Stealth build-it-yourself churn shows up there first. If your best customers are quietly leaving, the rest will follow once they see it works.
Don’t Build It. But If You Need To, Point Solutions Are The Easiest to Replace
We are not going to start building all our software ourselves. That would be insane and a massive distraction from running SaaStr.
But the boundary of where 90/10 lives has moved. Point solutions that have stopped evolving are increasingly on the wrong side of it.
We didn’t want to build a newsletter tool. We did it because the vendor froze in time, and 90 minutes a week of Amelia’s time matters more to us than 6 years of comfortable incumbency.
If you’re running a B2B SaaS product and you haven’t shipped anything material in a while, your customers may be quietly making the same calculation right now.
