The product worked. The business didn't.
After SISU I went all-in on something I cared about more than money — education. I'd spent twenty years watching the financial barrier between students and great teachers harden into permanent inequality. Plaito was built to dissolve it.
We shipped an AI tutor that worked across 40 languages and reached 1M+ downloads in 180 countries with 130k weekly active users. The product was loved. Students wrote us long emails about test scores, college acceptances, parents who finally understood algebra alongside them.
"We built it for the kid who can't afford a tutor. That kid loved it. That kid couldn't pay for it."
The economics never resolved. Inference costs at our usage levels outran our pricing in the markets we cared about most. We tried six different monetization paths. None of them honored the original mission while keeping the lights on.
I closed Plaito in 2024. It is the most valuable failure of my life. I am still going to fix this — just not yet, and not the same way. The next attempt will be hardware-aware, agent-native, and built on top of a cost curve that didn't exist when we started.