Should you be worried about Claude stealing your business?

Last week I built a teleprompter app for my Mac. This week, a Substack Notes scheduler. Each took about a day with Claude Code. Both do things the paid alternatives don't, or charge monthly subscriptions for features that don't need ongoing costs.

I'm not a developer. I'm a PM who finally has the tools to build exactly what I want.

Here's what's happening: the market for simple desktop tools has always relied on an 80/20 compromise. You bought software that did most of what you needed, accepted the missing 20%, and moved on. That compromise made sense when building your own would cost thousands or months of work.

Now it takes an afternoon.

The people building their own tools aren't random consumers – they're PMs, designers, and technical founders. Your most demanding customers, the ones who filed feature requests and noticed every rough edge. They're leaving first, and taking the feedback loop with them.

Not everything is equally vulnerable. Three things actually protect you:

Network effects – even small ones. The moment your product requires collaboration, it's not a weekend project anymore. Someone has to keep the database running.

Complexity – real technical depth that takes months to build properly, even with AI assistance.

Taste – a genuine point of view about how the tool should work. Something you can't generate from a prompt.

The floor is rising. What people can build for themselves gets higher every few months. If your only defence is "it works and it's convenient," you're relying on the 80/20 compromise, and it's getting worse for you every month.

The tools that survive will have something AI can't easily generate: a point of view, technical complexity, or collaborative infrastructure. Ideally more than one.

Read the full post: https://holenventures.substack.com/p/the-8020-compromise