the long version.

i started Savior in college, building a simple appointment-booking app for hospitals. then i spent real time inside them, watching how the staff actually worked, and the booking was never the real problem. the data was. patient records and schedules ran on paper and memory, and it broke all the time. so i changed direction and built software to run the hospital's whole operation instead.

the first version took three months and two pilot hospitals. it worked because it solved a painful daily problem instead of adding a nice-to-have. it cut the time to process a patient and got rid of the manual errors. after that it mostly sold itself. i went after chains like Fortis and Cloudnine for multi-location rollouts and grew from 35 hospitals in the first year to 85 by year three, with no sales team, all word of mouth. i built a team of 85 across engineering, operations and customer success, shipped a second version with analytics and decision support to push toward preventive care, and kept the company profitable without raising money. Zocdoc acquired us in 2020.

Career Leap started as a whatsapp group for people in tech who wanted to get better at the soft skills the job actually needs, learned from people who had already done it. weekly sessions with industry leaders grew the community from 2,000 to 3,500, and it hit 10,000 users in six months. then i built a learning app on top of it. it crossed 100k watch hours, and the gamification i added pushed session times up 50% and doubled retention.

when i talked to users, the real opportunity wasn't the app. it was live cohorts. with a team of three i ran two of them, trained 60 people at ₹10 to 15k each, and got 40% completion when most courses see 5 to 10%, because i built for the last lesson instead of the signup. i also tested a version for companies and ran it inside two of them. on every metric i cared about it was working. on the one that actually mattered it wasn't. the unit economics never held at scale, and no product was going to fix that, so i shut it down in 2023.

after building, i spent two years at Accel running Atoms, their pre-seed program for 0→1 founders. i ran it end to end, from picking the founders to sitting with each team through the hardest part of going from zero to one. four cohorts, two tracks, idea to first cheque.

backing founders taught me to read the early signal faster than i ever could as a founder myself. the teams that worked weren't the ones with the cleanest decks. they were the ones who got inside their problem the way i'd once tried to get inside hospitals.