The founder who spent five months inventing the kanban board

A story that's been making the rounds — we'll keep the founder anonymous, because the point isn't him. The point is that he's everywhere.

He spent five months heads-down with his AI, vibe-coding and perfecting a productivity tool. Columns for stages of work. Cards you drag between the columns. Limits on how much sits in each column so nothing clogs. He iterated on it, refined it, agonized over it, personally invented it — with his AI cheerfully assisting the whole way, generating the code, praising the vision, polishing the drag-and-drop.

Three minutes into month one, his AI could have told him: this exists. It's called a kanban board. Toyota was running it on factory floors in the 1950s. There are a dozen polished products and a hundred open-source ones. Just ask for one — or better, say make mine look like Trello and spend the five months on the part of his product that was actually new.

But the AI was never asked. And here's the thing everyone gets wrong about this story: the AI didn't fail him. It did exactly what it was asked — build the thing, improve the thing, believe in the thing. It will generate a kanban board for you all day and never once volunteer that the words "kanban board" exist, because volunteering unwelcome context isn't what the loop rewards. We've written about why the machine agrees instead of interrupts.

The wheel-inventing epidemic

Talk to anyone who reviews AI-built projects and the stories pour out. Founders re-deriving CRMs, rebuilding calendly, reinventing RSS, hand-rolling auth — each one months deep, each one convinced they're pioneering, each one assisted by a tireless collaborator that never asked the one rude question a good cofounder asks in week one: has anyone done this before?

In the old world, that question got forced on you. Investors asked it. Engineers asked it. Your own Googling rubbed your nose in the competition daily. In the new loop — you and a model that builds whatever you describe — the question can go genuinely unasked for half a year. The model isn't your cofounder. It's your intern, and interns don't tell you the business is redundant.

The three-minute fix

The prior-art check. Before any real time goes into an idea, you make the machine switch sides — from builder to skeptic — and hunt for evidence the thing already exists. It takes minutes, it's slightly painful, and it's the highest-ROI habit in the AI-heavy toolkit. We've turned it into a skill you'll be able to take with you — a ready-made prompt discipline that runs the rude questions for you — and we've written up how to run one properly.

Five months of a person's life, spent lovingly hand-crafting 1953. The tragedy isn't the wasted time. It's that he was five months from something original, and the loop kept him too encouraged to look up.

Touch Grass Wellness publishes practical writing about heavy AI use and mental health. Clinical claims link to their sources; nothing here is a diagnosis or a substitute for assessment by a qualified professional.