Precog draws an AI portrait of your future self at 1, 5, and 10 years out — built from your sleep, exercise, and water logs. Sunday at 7 PM, every week. iPhone, iOS 26+.
Hey Hunters 👋 Hyojin here, the maker of Precog. I built this after re-reading Hal Hershfield's Your Future Self (2023) and the 2011 Stanford retirement-saving paper. That study showed people who saw an aged version of themselves saved about 41% more for retirement — but only after a one-time VR session in a research lab. Precog is the recurring consumer version. Upload one selfie, log sleep / exercise / water through the week (one tap each), and every Sunday at 7 PM the AI draws your future-self portrait at 1, 5, and 10 years out. Three poles: best, prediction, worst. Three things I'm proud of: 1. No streaks. They burn out at month 6 and turn behavior into performance. We use a weekly aggregate viewing instead. 2. Privacy is genuinely careful. Apple's review made the consent flow better. The privacy policy is the most explicit Gemini disclosure we've seen. 3. The Sunday-at-7-PM cadence. It's the only honest viewing position — week ended, next not yet begun. Three things I'm uncertain about: 1. Whether the cadence catches on for users (year-1 retention will tell) 2. Whether the AI vivid enough to compound across 52 sessions/year 3. Identity preservation for diverse faces (mostly works, edge cases remain) The journal at https://precog.vibeland.app/blog goes deep on the research and the design decisions. Especially happy to discuss any of that today. What would you ask Precog at year 1?
Replies
Photos+
Hey Hunters 👋 Hyojin here, the maker of Precog. I built this after re-reading Hal Hershfield's Your Future Self (2023) and the 2011 Stanford retirement-saving paper. That study showed people who saw an aged version of themselves saved about 41% more for retirement — but only after a one-time VR session in a research lab. Precog is the recurring consumer version. Upload one selfie, log sleep / exercise / water through the week (one tap each), and every Sunday at 7 PM the AI draws your future-self portrait at 1, 5, and 10 years out. Three poles: best, prediction, worst. Three things I'm proud of: 1. No streaks. They burn out at month 6 and turn behavior into performance. We use a weekly aggregate viewing instead. 2. Privacy is genuinely careful. Apple's review made the consent flow better. The privacy policy is the most explicit Gemini disclosure we've seen. 3. The Sunday-at-7-PM cadence. It's the only honest viewing position — week ended, next not yet begun. Three things I'm uncertain about: 1. Whether the cadence catches on for users (year-1 retention will tell) 2. Whether the AI vivid enough to compound across 52 sessions/year 3. Identity preservation for diverse faces (mostly works, edge cases remain) The journal at https://precog.vibeland.app/blog goes deep on the research and the design decisions. Especially happy to discuss any of that today. What would you ask Precog at year 1?