Everyone's telling you to put AI into everything. This episode is the counterweight. Paul brought back paper sign-in sheets at his martial arts academy after turning down a "we can vibe-code you an app in 75 minutes" offer, and student testing tripled in eight weeks because the paper created a ritual and a community touchpoint an app couldn't. Ryan explains why an AI avatar was the wrong call for a human-first channel, and why a consulting client's customers specifically wanted a real person. The throughline: ask should you, not just could you. Protect the human parts of your business on purpose, and point AI at the back office where it actually earns its keep.
Risk rating: 🟢 Safe to start (this is a decision-making lens, not a tool to install)
If you take one thing from this episode...
1. Make a two-column list: where a human IS the product (keep it) and where the work is invisible to the customer (automate it).
2. Run the should-you-versus-could-you test before adding any AI. "Could" is almost always yes. "Should" is the real question.
3. Never bolt beta or half-ready AI onto a process your business can't run without. One break can cripple a small shop.
Poll: Where would you NEVER let AI take over in your business?