⏱️ Read time: ~2 min

A client called last week. We're between contracts, but their AI chatbot subscription still runs on my account. They asked if we could keep it that way until they build more internal AI capabilities. I said yes, but added: "Please don't expect a monthly invoice. It's only around 400 euros, so the admin overhead isn't worth it."

Their response: "But you could automate that billing, right?"

True. I could. A quick Zapier integration, maybe 30 minutes of setup. Totally doable.

I said no.

The Automation Trap

Here's what I've noticed: AI doesn't necessarily make you more productive. It often makes you build more things. And most of those things don't move the needle.

The chatbot billing is a perfect example. Automating it is possible. But it's not worth it. Here's why:

This is a one-off favor, not a repeatable pattern. It's an exception, not a system. If I automate it, I'm spending 30 minutes of setup time to save myself maybe 5 minutes of manual work, once or maybe twice. That's a terrible ROI.

But because AI (and no-code tools) make automation so frictionless, the temptation is real. You see a repetitive task, and your brain goes: "I could fix that in minutes." So you do. And suddenly you've built a system for something that didn't need one.

The Real Skill

The actual skill isn't "how to use AI." Nor is it "can AI do it?". It's "what should I actually build?"

That requires brutal honesty:

• Is this repeatable, or is it a one-off?
• How much time will setup take vs. how much will it save?
• Am I building this because it matters, or because I can?
• What's the real cost of maintenance and future changes?

For the chatbot billing: setup cost (30 min) > quarterly savings (5 min) × once or maybe twice. Not worth it. I'll do it manually when it comes up.

For something else, say, a weekly client report that takes 45 minutes and happens every single week, an automation could be a no brainer.

Key Takeaway

Productivity gains are real, but only if you ruthlessly prioritize what's worth building. AI doesn't solve the prioritization problem. It makes it harder, because now you can build almost anything in minutes. The skill is knowing when to say no, even when it's possible.

Self-leadership totally beats automation hacks every time.

Keep experimenting

Elena

P.S. What's one thing you've automated recently that you now regret? Or one thing you could automate but haven't, because the ROI doesn't make sense? Hit reply and let me know what the real-world math looks like for you.