I built a customer care bot that can put a price on anything.
It fails at exactly one thing.
I work on AI for Sales and Service. So for a side project I built the most „efficient“ customer care imaginable. The one every KPI dashboard secretly dreams of.
Its only job: tell you what something is worth. Your grandmother’s watch. Your morning coffee. Your brother.
It answered every single one. Politely. Instantly. With a number.
Here’s the „Interesting“ part: for a change there’s no LLM behind it. No model, no API.
A few hundred lines of JavaScript and a pricing table. The eerie thing about „care“ that only knows how to value isn’t the tech. It’s the design brief.
And yet there’s one thing it can’t price. I didn’t hide it well. It just doesn’t occur to people to ask.
AS much as I like to experiment with the fringe there are still some things I’m taking back into real service design:
• If your bot answers everything, ask what it’s optimizing for
• Warm language on top of a cold objective is still a cold objective
• The moment a customer becomes an asset, you’ve stopped doing service
So, here ya go! Here are the rules:
Try to break it!
Don’t post what worked.
Post what it offered you for the thing you care about most.
Link’s in the first comment.
#Fiction #Experimentation #CustomerExperience #AI #ServiceDesign #ARG
