1998The co-founder who already did it. Shekhar had been in the Valley since 1991 and was part of a company that went public in 1998 at a couple of billion dollars, then sold on the following year for around four. He later sold a second company to Juniper and now sits in Bangalore. He knows how to sell and make money, which Prasanna rates above knowing the market.
2005-11Prasanna makes the mistakes himself. Two startups over those six years, both B2B, the second of a kind we would now simply call SaaS. Upekkha's curriculum is largely a list of what went wrong in them.
2008-09The ecosystem is a handful of rooms. Barcamp, Product Council and Unplugged were about all there was, and Rajan was already building AI startups, very early, while running Mobile Monday. Websites and forums, no groups.
2010Freshworks gets covered, before the smartphone. Vishal was writing about the company that year, when hardly anyone carried more than a BlackBerry. India is called the SaaS capital now, and people in Chennai will tell you Chennai is.
2017Upekkha opens. Curriculum, community and cohort, with a cheque of about 100,000 dollars a startup, built to save founders the two or three years Prasanna lost. The first three or four years went on working out what actually helped.
2020Indian SMBs graduate from Tally. Most had bought no software but Tally, and only started buying anything else around COVID, against 40 or 50 years of software buying in the United States. Which is why onboarding a customer is cheaper in America, not dearer.
2023The sand shifts mid-recording. A data migration that took a developer days now takes minutes, an analytics founder tells him the state of the art beats the analysts she trained, and one infrastructure company that everyone in India has used has stopped hiring junior developers. The 10 million dollar fund is fully deployed across roughly 120 startups, three of them past 5 million dollars of revenue.
Sep 2023The next cohort. Twenty-five to thirty startups chosen from about 400 applications, with applications closing at the end of June. The plan is to take more per cohort without adding people, using GPT and automation instead.
