All plays › Shops and e-commerce

86’d · What not to do
Shops and e-commerceGrok Bot

86 this: a safety check that warns but still passes

Picked and checked by Olga Pearson, host of The Practical AI Show · Published

What happened

self-reported

A late-package check passed while an order sat 27 hours with no proof it had shipped, because four orders were saved without a sale time. The bot’s nightly review caught it, and the rerun failed and named the late order.

Ryan Hanley runs a sports card business on the side with Grok Bot. One bot’s whole job is to flag late packages. But four orders had been saved without the time they sold, and a fix from the night before, meant to stop the check from crashing, turned that missing time into a warning. So the check warned and still passed, while an order sat 27 hours with no proof of shipping. His bot Carl caught it in a nightly review and changed the rule: if the sale time is missing, use the time the shipping label was printed. If there’s no time at all, the check fails. The rerun failed and named the late order.

Who did it

Ryan Hanley (@rhanley)Sports card business (side business)Ryan Hanley on X · Oct 25, 2026

Try it now

The safe way to do this instead:

  1. List the checks your bots run for you (late orders, unpaid invoices, missed follow-ups).
  2. For each one, ask: what happens when a piece of data is missing?
  3. Make missing data fail the check, not just warn, unless there’s a safe backup (like the label print time).
  4. Have a bot review the day’s checks every night and look for passes that shouldn’t have passed.
  5. After any fix, rerun the check on a known bad case and make sure it fails.
Copy-paste prompt
Review this automated check: [describe it or paste the rules]. List every way it could pass when it should fail, especially when data is missing or blank. For each, tell me the safer rule (use a backup field, or fail loudly). Then give me one test case that should fail, so I can confirm the fix works.

Make it run every week

A play works once. Turn it into a routine in three small steps:

  1. Run it once and check the result yourself.
  2. Give your AI a schedule. Tell it when to run this play (every week, or every month) and to message you with what it found before you act.
  3. Keep a log. One line per run: what was checked, anything flagged, what you did.

See how owners turn plays into systems →

What to watch out for

Self-reported. The lesson: a crash fix that turns errors into warnings can hide the exact problem the check exists to catch. Warnings get ignored; failures get fixed.

Did it work?

PageMotor keeps this site free.