2025 Annual Memo
I have a habit of writing annual memos personally, but I rarely publish them. They usually end up like all the other stuff I write internally about business, psychology, important decisions I make, etc., collecting dust in my Notion and Notes.
Writing has always been a way for me to think, debug, and understand my thoughts better, and 2025 deserves a public review. It has been a pivotal year for me (the last big pivotal year for me was 2021). Here are some reflections on the year and the heuristics I developed in entrepreneurship, startups, and life in general.
The Journey So Far
Before I review 2025, here's a quick context on my background.
For those who don't know me: I am 31, live in Bangalore, and recently got married. I was raised by a single mom in a house of five that operated on a cash flow of less than $4k a year. Living well below our means until I started earning at 21 wired me for survival. It taught me the importance of operating under constraints and being resourceful — lessons that naturally transferred to navigating the chaos of building startups.
Over the last 10 years, I've been fortunate enough to have a driver's seat view of building multiple startups. In almost every case, whether as a founder or founding member, I entered at the -1 to 0 stage. Spending a decade in this zone has helped me build a specific set of mental models one needs to navigate uncertainty, cross the -1 to 0 chasm, and subsequently the 0 to 1 chasm.
One of the important lessons I've learnt is that timing is the ultimate leverage. You are too ahead of time, and you end up spending years creating the market — you are too late, and you end up playing cat and mouse with other startups and incumbents. The sweet spot to build something is when you have a contrarian hypothesis on a market with tailwinds that seems the right dose of contrarian enough, which potentially unlocks a large market or changes the status quo if one succeeds, and 2025 was the year when I finally saw this window of opportunity again.
2025: The Year of Conviction
2025, in hindsight, has been one of the big +EV years for me. There are a bunch of commitments I made this year that are irreversible, got back to founder mode again (and boy, I can't tell you how alive I feel again), embarked on a direction that can potentially change labour markets, and most importantly, got married to the love of my life.
Here are some highlights for the year in no particular order —
Deep Reading & Learning
Read ~30 books this year. I started this year reading 1 book a week. I was very much immersed in learning mode at the beginning of this year. It was a habit that I cultivated at the end of 2024, and it continued into 2025. Some books that I'll definitely reread in 2026 and highly recommend:
- Metamorphosis of Prime Intellect
- Poor Charlie's Almanack
- What Life Should Mean To You by Alfred Adler
- Born of This Land: My Life Story
P.S. The habit kind of broke when I got into full-time founder mode towards the middle of the year. Going to bring this habit back in Q1 again and target reading more in 2026 — more sci-fi, more history, and more of things that pique my natural curiosity.
Lossfunk & Research-First Culture
Joined Lossfunk and got exposed to research-first culture — thanks to Sudarshan, who was in batch 1, became a good friend, and eventually introduced me to Paras. For those who don't know, Lossfunk is an AI research lab led by Paras Chopra, which is primarily a place for researchers who want to do good science from India.
I read a book in 2025, Where Good Ideas Come From by Steven Johnson, where he talks about how environments become fertile grounds for idea formation — that the act of creation rarely happens in isolation but emerges as dots that connect in our minds as networks, emerging from slow hunches to conviction. Lossfunk provided the fertile ground for me to wander, feed my curiosity, and let ideas riff off each other. I've said this a bunch of times to Paras directly, but I owe a huge debt to Paras and Lossfunk for helping me discover my new north star.
Exploring LLM Psychology
Worked on LLM psychology and built something at the intersection of Safety x RL environments x Social Intelligence. The core idea here was that as LLMs start generalising, it'll be important for us to start understanding their psychology. So we tried to put language models in constrained environments to test cooperation vs greed and built evals around them. This is probably not venture-scale business, but important work as an adjacent space of AI safety. If you or someone you know is working on this, my DMs are open, and I'd love to exchange ideas, notes, and learnings on this.
India, China, and the US
Spent a week in Koh Tao and got my first advanced water certificate, did my first trip in China, visited Las Vegas, and made multiple trips to SF.
This was also my first visit to China, and it solidified my conviction that 3 countries will define the trajectory of this century — India, China, and the USA.
There's a lot for India to learn from the manufacturing velocity of China and the innovation and technological acceleration in the USA, especially in AI and Robotics. We cannot afford to sit this one out. India missed a bunch of technological innovations over the last 70 years because India didn't have the fertile ground to flourish — case in point: semiconductors, computers, and the development of the internet (all of which happened prior to India getting liberalised). Post liberalisation, we took some time to catch up, but the last 20 years have changed the ground reality for us. There is plenty of risk capital at stake, a more mature ecosystem, and founders aiming for more ambitious ideas. We have the ingredients, and we must not sit back on this race.
Got Married
Got married to the love of my life. This has been the single biggest arc in my journey. She's been my biggest cheerleader and support system, taught me how to put family first and love animals unconditionally, and is a constant reminder of what it means to be empathetic and kind in this world.
New Friends & Family
Made great new friends in 2025 (if you are reading this, you know who you are) and have a much larger family with the addition of my in-laws. This is a stark difference between this year and last. I was in Bangalore last year, but was kind of homebound — I had a small 3-month-old pet "Kaaza" with separation anxiety, so most of my time was spent locked in the house and doing remote work. 2025 was the opposite — Kaaza got older, solved his separation anxiety, we got a dog sitter, and I started stepping out more often. Key learning: step out, make friends, touch grass, and do not be locked away in the house at any given point. Do not fall for the remote work trap. It strips away the serendipity of meeting new people who might become future cofounders, friends, partners, mentors, or peers.
Back to Founder Mode
Back to founder mode in the second half of the year, to start a company to push the boundaries of general-purpose robotics from India.
I started working on this idea around August/July 2025, but the roots go back to when I exited my previous startup and was very clear that the next company I build will be an idea I can commit to for 10 years and will be missionary and not mercenary. I've been fascinated by labour markets and have a first-hand understanding of how technology shapes them.
Robotics as a field is 70 years old, but only lately has the industry started taking off after we reached critical inflection points in language and video models. Robots today have a body but are limited by their understanding of the world. We have made some headway here over the last 6 months and are excited to come out of stealth mode and talk more about what we've been cooking behind the scenes.
The Team
Put together a special team who are equally committed to the above idea as me. I feel very grateful and lucky to be working with people who inspire me every day to come to work and push the boundaries of general-purpose robotics from India.
Heuristics & Takeaways
Some takeaways that apply to business and personal life, which were consolidated in 2025:
- Permissionless agency is scarce. We spend the first 20 years of our lives in traditional institutions (schools, entry-level jobs) designed to make us compliant and punished if we deviate. We are taught to respect permission — to adhere to the rules — to not bend them. But in a world increasingly dominated by AI, you have to unlearn the above. It took me the last 10 years to unlearn everything I was taught in school — there are no teachers, no permissions, and no rulebooks to play by. You don't need a degree to do research; you don't need a title to lead; you don't need funding to start building. The world is much more malleable than we think. Stop asking for permission to do the work you want to do. Just do the work, and let the permission catch up later.
- Always take contrarian bets. It seems obvious in hindsight, but the only structural advantage a young founder or team has against an incumbent is the ability to be wrong. Big companies naturally converge on conventional bets as they derisk themselves over time and focus on the protection of value. Early teams focus on value creation and not protection, and therefore, they need to take more contrarian bets to stand out against the incumbents.
- Take more risks and aim for bigger swings. This doesn't mean being reckless or gambling your money away — it means refining your decision-making engine. A good mental model here is Expected Value (EV), a concept borrowed from poker. In life, most people default to minimizing loss (loss aversion). They avoid any bet where there is a chance of failure, even if the potential upside is 100x. But the best operators think in +EV. They look for asymmetric bets: opportunities with capped downsides (you lose a set amount of time or money) but uncapped upsides (you change the industry or your life trajectory). All you need to do is take a couple of these +EV bets to change your life trajectory. You won't win 100% of the time, but you need 1 win that makes the rest of your career look like a footnote.
- Don't just build startups for the sake of building. Build projects and wander around them before you commit to an idea. I made a classic mistake in 2023 when I committed to the first idea that appeared after I exited IndiGG — I fell for the void of starting the next thing just to feel the motion of building again. I put together a team and started executing before I built conviction in the idea. Big learning from this is to have a clear distinction between wandering and execution. Think of wandering as breadth and execution as depth. Get as much breadth as possible, work at or consult startups, build multiple projects, and stay lean for as long as you can. Startups cannot be built until you fully commit to yourself and are ready to get punched in the face for years without seeing the light of day. Commitment requires conviction, and conviction requires exploration.
- Be ruthless about your decision once you've made up your mind. Take time to think, but once you make up your mind, act decisively. Don't put yourself in 2 boats. Commit to an idea or person regardless of whether it's easy or hard, and do whatever it takes to stand by your decision once you commit.
- Make good friends and stand by them when they need you. It takes commitment, like all other things, to keep a friendship, but a single moment to break it. Accept that good friendships come with maintenance costs like all other things in life.
- There has never been a better time to build in the history of humanity. If we zoom out, the history of progress in humanity involves reducing the cost of fundamental inputs. The Agricultural Revolution reduced the cost of calories. The Industrial Revolution reduced the cost of physical labor and energy. The Internet Revolution reduced the cost of distribution. We are now living through the Intelligence Revolution, where the marginal cost of cognition is trending toward zero. There has never been a better time to build in the history of humanity.
Happy New Year, and see you in 2026!
← Back to home