JTBD Statement Checker

Role
Sole designer + builder
Timeline
April 2026 | ~10 hours
Core skills
  • Jobs-to-be-Done domain expertise
  • AI-assisted building
  • Visual + interaction design
  • Shipping + public launch
Collaborators
Just me + Claude
Built with
  • Claude
  • GitHub
  • Vercel

An AI-built tool that judges whether a Jobs-to-be-Done statement is the real thing, cites the principles in play, and suggests a rewrite. Shipped solo in about 10 hours.

Why this exists

If you’ve worked with the Jobs-to-be-Done theory before, you know how powerful it can be for understanding your audience, and how hard it can be to get right.

At best, I’ve seen well-intentioned teams struggle to write accurate JTBD statements. This is expected — it’s harder than it looks. It requires a mindset shift: removing yourself and your product from the equation entirely, which is antithetical to the way most people and companies operate. It’s an exercise that takes focused practice.

At worst, it’s become the latest corporate buzzword. I often see the “JTBD” label slapped on go-to-market briefs and acquisition targets…the exact kind of company-first thinking that JTBD is designed to move you past.

We’ve lost the plot a little when it comes to this theory. So I built a tool to help steer us back on track.

Try it out

How it works

You paste in a statement (or drop a screenshot of one), and it tells you whether it’s a genuine job-to-be-done, partway there, or not a job at all. It explains its reasoning, cites the specific JTBD principles in play, and, if the statement needs work, suggests a rewrite.

Whether you’re actively trying to get JTBD right on your team or just curious whether that statement in your strategy deck is actually a job…give it a shot. Sipsy the Milkshake is waiting on the other side to guilt trip you with puppy eyes for moral support.

Try it just for the goops

I spent an embarrassing amount of time (and tokens) prompting in circles to get Sipsy’s animations right. I never thought I’d spend part of my work day telling an AI to “make the milkshake drips goopier.” But they are goopy. And they deserve attention.

Try it out