beehiiv CEO Tyler Denk Is Right. Vibe-Coding Is Short-Term Thinking.

beehiiv CEO Tyler Denk says someone tells him every week they'll vibe-code their own beehiiv over the weekend. His post on X lists what they'd also need: a 99.3% delivery rate, infrastructure to send 3 million emails a minute, bounce handling, blocklist monitoring and bot filtering. He's right. Building the app is the easy part. Running it for years is the business.

I've spent almost 20 years in software, and the companies that lasted were the ones that played the long game. AI hasn't changed that. It has made the short game look easier than ever.

I'm a fan of building with AI. I moved this whole site off WordPress and rebuilt it that way. That's exactly why I take Denk's list seriously.

Can you really vibe-code beehiiv in a weekend?

You can build something that looks like it. A signup form, an editor and a send button. That part is easy now, and I love that it's easy.

What you can't build in a weekend is what makes the email land. Gmail and Yahoo now require one-click unsubscribe. Every bounce has to be sorted into try again or never again. New domains need warming up on a schedule. Bots sign up and click links, and someone has to filter them out of every report.

None of that shows up in a demo. All of it shows up the first time your newsletter lands in spam for 40% of your list.

What happens after week one?

Nick Velkovski, a SaaS founder, put it plainly on X: "SaaS founders don't really get paid to build the product. We get paid to keep it running." His list of what week two looks like is one every operator will recognize.

• A dependency ships a breaking change and the build fails for no reason.
• An SSL certificate expires.
• A database change has to run while customers are using it.
• Payments fail quietly, so someone paid and didn't get access.
• A privacy deletion request means you need to know everywhere a user's data lives.
• A bigger customer sends a 200-question security review.

Every item is boring. Every item is also a reason customers stay or leave. This is the work that never ends, and it's the work AI makes easy to forget about when you're excited about the first version.

Isn't software headed to zero?

The cost of writing code is. I've written that legacy SaaS is worth its data and that Bending Spoons runs big products with small teams. I still believe both. Code is getting cheap. Running a product well for years is not.

That's where the value moves. Trust with mailbox providers. Years of data on what works. A team that picks up the phone at 2 a.m. You can't vibe-code a reputation. You earn it over time.

beehiiv's real product isn't the editor. It's the years of relationships with Gmail, Yahoo and Outlook that get your email into the inbox. That's the moat, and a weekend build doesn't have one.

Why is vibe-coding short-term thinking?

Because it measures the wrong thing. A weekend build measures how fast you can launch. A business is measured by how well it runs in year three, after the excitement is gone and the edge cases have piled up.

Saving $50 a month on a newsletter tool feels like a win. Then you spend 10 hours a month keeping your own version alive. If your time is worth more than $5 an hour, the math was never close.

The same thinking shows up inside companies. A team builds an internal tool in a week with AI, nobody owns it, and a year later it breaks during the busiest week of the quarter. The build was fast. The ownership was never planned.

When should you build instead of buy?

Sometimes building is right. If the tool is the core of what makes your company different, own it. If it's plumbing everyone needs, buy it from someone who lives and breathes it.

Before you build, run it through a few simple tests.

1. The week-two test. What does this cost to run for five years, not to launch in one weekend?
2. The 2 a.m. test. When it breaks at night, who gets the call? If the answer is nobody, don't build it.
3. The core test. Is this what customers pay you for, or is it plumbing?
4. The reputation test. Does a vendor like beehiiv have trust you'd need years to earn?
5. The hours test. Multiply the hours to maintain it by what an hour of your time is worth.

How do you play the long game with AI?

Use AI to go faster on the parts that make you different. Let the experts handle the parts that don't. Spend the time you save on customers, not on maintaining tools nobody asked you to build.

If you want a fast read on how exposed your own company is, take the AI-proof test. It takes a few minutes.

The weekend build is a great demo. The long game is what's still running in five years.

Disclosure: I don't own beehiiv stock and have no business relationship with the company.