Bets Log
Timestamped predictions. Updated when the market confirms or refutes.
I make public bets about technology, markets, and India’s tech ecosystem. This page timestamps them so I can’t hide from being wrong.
I’ve been making bets like these with my career for eighteen years. The historical section below is the scorecard — right calls, wrong calls, and bets I chose not to take. The active bets are the current ones.
Format: Each bet has a date, a thesis, and an outcome. Confirmed and wrong bets link to the evidence. I don’t edit bets after they’re made. I only add outcomes.
Active Bets
Classical software delivery will be commoditized within two years
The current model of building, deploying, and running software — CI/CD pipelines, container orchestration, observability stacks — will become a commodity layer. The new frontier is agent workforce delivery and orchestration: deploying, monitoring, and governing autonomous AI agents at scale. The companies that figure out agent reliability infrastructure win the next era.
Skin in the game: I’m building Zopdev on this thesis. Zopday (PaaS in your own cloud) handles the delivery layer. Zopnight (cloud governance and cost optimization) is gaining more traction — which itself suggests the delivery layer is already becoming table stakes.
AI agents will have a reliability crisis before a capability crisis
The bottleneck for enterprise AI agent adoption won’t be what agents can do — it’ll be whether they do it consistently. Indian SaaS companies building reliability infrastructure will capture disproportionate value.
Related: Why Indian SaaS Companies Will Win the Agent Reliability War
AI runtime infrastructure becomes the new cloud infrastructure play
The companies building inference optimization, model routing, and GPU orchestration layers will be as valuable as the cloud providers themselves within 5 years.
Related: AI Runtime Infrastructure Is the New Cloud Infrastructure Play
LLM-as-judge won’t work for Indian market evaluation
LLM judges trained on Western corpora will systematically misgrade Indian language, cultural context, and domain-specific outputs. Teams that ship LLM-as-judge without local calibration will get burned.
Related: The LLM-as-Judge Stack Is Dead on Arrival for Indian Markets
Federated learning won’t save Indian healthcare AI
The federated learning pitch for healthcare — “keep data local, train globally” — will fail in India because the problem isn’t data privacy regulation. It’s data quality, hospital IT infrastructure, and misaligned incentives.
Related: Why Federated Learning Won’t Save Indian Healthcare AI
Most corporate training will shift from certification to outcome-verified skill development by 2028
The majority of corporate training budgets will move from “courses completed” to “can this person actually do the task.” Not certificates issued — measurable execution ability.
Related: Building an Upskilling Platform Before COVID Made It Obvious (forthcoming)
Offline-first communities will outperform digital-first on retention
Communities that start with physical gathering and layer digital on top will have 3-5x better 12-month retention than communities that start digital and try to add IRL events.
Related: The IRL Growth Bet (forthcoming)
Historical Bets
I’ve been making these bets for years — they just weren’t public. This page makes them so.
K12 ed-tech will be a real market in India
Indian schools will adopt technology for curriculum delivery. There’s a market for digital learning products aimed at K12 students.
What happened: Confirmed — but I was early by about five years. Built Digital PACE out of college, sold it. The funded ed-tech wave (Byju’s, Vedantu, Toppr) arrived around 2013-2015. Being right on the thesis and too early on the timing is a pattern I’d repeat.
Performance arts and dance will become a viable career path in India
Dance and choreography in India is about to break out of the “hobby” box. Reality TV, Bollywood demand, and cultural shifts will make it a legitimate, well-paying career.
What happened: Confirmed — but I chose not to take this bet. Founded InSync, a dance club at IIT Bombay. Led teams to multiple competition wins. Was selected for a Bollywood movie (turned it down — would have meant dropping a semester). Worked with director Sudhir Mishra on projects post-college. Made the top rounds of Dance India Dance. Within a couple of years, dancers became household names, dance emerged as a real career, choreography became a respected profession. I chose tech instead. Not a wrong bet — a bet I deliberately didn’t take.
Leisure experiences are an underserved marketplace category
People want to discover and book unique experiences — dance classes, adventure activities, cooking workshops — the same way they book movie tickets. This is an unserved marketplace.
What happened: Confirmed on thesis, wrong on execution and timing. Built Tushky — India’s first leisure experiences marketplace. Isango and GetYourGuide existed internationally, but nothing like this existed in India. Raised from 500 Startups (Batch 6, Spring 2013) and GSF India. Airbnb launched Experiences in 2016, five years later. The market was real. Our ability to build a sustainable business in it at that time was not.
Creator-powered commerce needs a three-sided marketplace
The future of merchandise and fashion isn’t brand-to-consumer or manufacturer-to-consumer. It’s a three-sided marketplace: manufacturers on one side, creators (photographers, graphic designers) on another, customers on the third. The serious creators will launch their own D2C brands.
What happened: Partial. Short stint at FreeCltr, but the thesis mapped almost exactly to the D2C wave that followed in India (2017-2020). Creators did become brands. The marketplace model for enabling it didn’t win — vertical D2C platforms and Shopify-style tools did instead.
Marketplaces need to control inventory to scale, not just aggregate it
OYO’s aggregation model (listing hotels without controlling the experience) has a ceiling. To deliver consistent quality at scale, the company needs to own or deeply control the inventory — not just list it.
What happened: Confirmed. As Product Head at OYO, I led the marketplace-to-owned-inventory pivot. OYO’s subsequent growth inflection followed this transition. The broader lesson applies beyond hospitality: aggregation marketplaces that don’t eventually control quality hit a wall.
PM upskilling will become a mass market
Product management is becoming a recognized career path in India, but there’s no structured, practical training for it. Professionals will pay for outcome-oriented PM education — not certificates, but real career transitions.
What happened: Confirmed. Started Pragmatic Leaders with lightweight bootcamps in 2017, ran the first formal cohorts in 2018, scaled from 2019. Trained 10,000+ professionals. YC W21. COVID massively accelerated the upskilling market, but we were already operating at scale before the wave hit. 67% of early students transitioned to product roles at companies like PayPal, Hotstar, and Paytm with salary hikes up to 300%.
Scoring Rules
- Confirmed: Market evidence validates the thesis. I link to the evidence.
- Wrong: Market went the other direction. I explain what I missed.
- Partial: Directionally right but magnitude or timeline was off.
- Pending: Not enough time or evidence to call it yet.
- Didn’t take: I saw it, I could have committed, and I chose a different path.