Since college, I’ve dreamed of starting my own tech company. In early 2022, I left a job I loved at a unicorn, surrounded by the most impressive people I’ve encountered, to chase this dream.
I initially thought it would take me just three months after leaving my job to figure out what I wanted to build for the next 5-10 years. It ended up taking about eight times longer. I had been fascinated with web3, working on related projects during my nights and weekends since 2017. So, in 2022, I went full force, traveling around the country to build, ideate, and meet potential customers and collaborators, while witnessing daunting events like the Terra/UST death spiral, multiple hacks on exchanges and token bridges, and ultimately, the dramatic downfall of FTX.
As 2022 drew to a close, my long-standing interest in renewable energy reemerged, particularly triggered by a team/concept around a token-incentivized, decentralized virtual power plant. I had always envisioned venturing into clean energy after starting a company or two. However, the challenging environment of building a consumer-facing web3 company prompted me to pivot — while I remain optimistic and passionate about the web3 space. This shift in focus led me to immerse myself in clean energy and, eventually, applications of AI in this sector.
This past year, 2023, involved brute-forcing my way into an industry I had never worked in before. I developed my intuition through conversations, hands-on work, building tools, aggregating data, and traveling around the country to meet potential customers and advisors.
On a more personal note, this has been the hardest thing I’ve ever done. I’ve faced more rejection in the past two years than in the rest of my life combined. From minimalist living and multiple relocations to attending a close friend’s funeral and another’s wedding, it’s been quite the journey. I’ve been a collector of accolades for most of my life: an award-winning student in high school, an Ivy League education, engineering/leadership roles at Microsoft, Apple, and Brex, scholarships from Google/Facebook/Square, hackathon prizes and selective fellowships. But the startup journey was a huge, humbling slap in the face (with more punches and slaps forthcoming). For those interested in my personal anecdotes, I’m open to sharing more one-on-one. 🙂
The following post is an objective summary of the things I built, wrote, and learned over the last two years. I did ultimately find a path in software and energy that I’m thrilled to unveil in 2024, with a strong team. As with the last two years, there’s still a lot of risk and uncertainty. But my skin is a bit thicker now, my pain tolerance a little higher, while I remain as starry-eyed as ever.
2023 Outputs
Things Built
SolarGPT (August - Django, React, Pinecone, Langchain): I noticed that a lot of developers of clean energy projects had questions about the Inflation Reduction Act (especially as it relates to tax credits). So I thought, why not put LLMs to use and let people ask questions to it? I taught myself the chatbot stack and threw a bunch of legal docs into a vector database. By using my existing mailing list and contacts, I was able to launch user trials for SolarGPT and start collecting data on the questions asked by clean energy project developers. 500 unique visitors came through word of mouth
Substation Search Tool (Dec - Django, React, Next.js, Render, Postgres): A recurring pain point from developers was substation (distribution) level data. I noticed that each utility or grid operator had their own website and GIS tool to showcase hosting capacity. I collaborated with Steve and Tae during the holidays to aggregate this data from various utility websites. We quickly launched the tool to some design partners and a mailing list
Community solar search tool (Jan - Django, React, Next.js, Postgres): To capitalize on the community solar trend, I scraped and aggregated data from around 8 lead-generation platforms for solar customer acquisition, creating a useful tool for developers in this space. This is one of the most comprehensive catalogs of community solar projects in the country, with 300 projects and counting
…and more prototypes and figma demos in stealth
Ideas Tested
- Fractional ownership/investment in clean energy projects (Jan-March)
- Tax credit marketplace (April-May)
- Asset marketplace + virtual data rooms (June-July)
- DoNotPay for Brazil (August)
- Residential solar leadgen (October)
- Current idea/company (in stealth, September onwards)
Media/Publications
- Climate + LLMs (2k reads)
- Project Finance 101 (2.3k reads)
- Django cheatsheet (5k reads)
- Tax Credits (featured by Medium, 700 reads)
- Avoiding a Cleantech Bubble
- Fall of a Climate Bank
- IRA guidance
- Energy developer Resource List (100+ developers accessed)
- TED-style talk about LLMs in energy project development (starts at 48:00)
- Dream Big Library (compilation of inspiring content about clean energy)
- The Grid Reading Notes
- Newsletter with 120 prospective customers (stealth)
- My all-time most popular tweet, thanks to the GPT craze
People, Fellowships & Events
- Recruited 3 energy industry veterans as advisors (Stanford grads, exited to SolarCity, SVB and British Petroleum Lightsource leaders), from 10+ candidates
- Hundreds of customer research conversations and demos
- 7 cofounders considered/trialed
- 2 conferences: RE+ Community Solar (San Diego, January), RE+ Main Event (Vegas, September)
- Clean Energy Leadership Institute (Fellowship, Aug - December)
- Raised from exceptional angels
2022 Outputs
Things Built / Ideas Tested
Lua, No-code Token Minter on Polygon (#1 Product of the Day on Product Hunt): Ethereum transactions were slow, and Layer 2’s were having a moment. I wanted to test out the developer experience on Polygon, a Layer 2 protocol that processes 65,000 transactions per second (compared to Ethereum’s 16 TPS). It ended up getting some traction through PH and I had to deal with some support requests from people trying to use it for very unconventional token drops…
Trophy Case (NFT Hack, winner of Polygon prize): Grace, Melanie and I teamed up for NFT Hack, where we had the idea of minting badges one collects from various different apps and games on chain so that you can have them all in a universal wallet
Bridgedex: 3rd place at Jump hackathon in Chicago, $10k prize. I spent about a month in Chicago as part of an extended hackathon organized by Jump Trading. I noticed a proliferation of cross-chain bridges which allow users to “bridge” a token from one chain to another. Ever the aggregator, I made a simple catalog for anyone to figure out the bridges they can use to take a token from chain A to chain B. I was building for other crypto-native folks and “degens” so the hackathon was a good opportunity to build, talk to users, and iterate
Optimistic Cafe (ETHNewYork, winner of Uniswap and Optimism prizes): Brought together my talented friends Justin, Romain, Akilesh and Daryl during a summer in NYC to play with the Uniswap and Optimism SDKs
Terragram (Miami Hack Week): Send gift cards with stablecoins
Crypto payments & invoicing tool: A system allowing global freelancers to invoice in their preferred cryptocurrency, simplifying cross-currency transactions
Crypto payouts: A solution for e-commerce businesses in economically volatile regions to convert their earnings into cryptocurrency, mitigating currency depreciation risk
Token-incentivized energy demand response: A token-based reward system for individuals reducing energy consumption or contributing clean energy to the grid
Media / Publications
- MetaMask blog post (8k reads, got translated into Chinese)
- MPC blog post (3k reads)
- Brex/why join a startup blog post (7k reads)
- Django cheatsheet
- Graded (High School) Alumni spotlight
- Blockchain bridge explainer thread
- Use crypto to support ukraine thread
- Cross-chain bridges thread
- Wormhole explainer thread
- How crypto wallets differ from bank accounts thread
- Crypto + climate thread
- Summary of Matt Levine’s post on crypto thread
People, Fellowships & Events
- Crypto EIR with Human Capital and Bain Capital Ventures
- SPC Founder Fellowship offer
- Judged HackMoney alongside Stani Kulechov, founder of Aave
Lessons
- Be decisive. Making decisions and taking action quickly is better than trying to over-optimize at the cost of action
- Pick a side/decision and own up to it. When you try to make everyone happy, you make nobody happy
- You don’t have to respond to every message or email. But for the ones that matter, respond as quickly as you can
- Always overprepare for all important conversations. Doing your homework goes a long way
- People-pleasing and the inability to prioritize ruthlessly/give transparent answers will kill you
- Take the time to zoom out instead of jumping into the next thing out of fear and scarcity mindset
- There is no such thing as the perfect idea. Know your non-negotiables and when enough of the boxes are checked, just go for it. At the same time, keep the bar high for the things that matter. As Jobs says, “As with all matters of the heart, you’ll know when you find it.”
- For a cofounder-type partnership to work (in work and in life), there needs to be chemistry, equal commitment, and alignment in intentions. This should take precedence over credentials, raw intelligence, and other forms of shininess
- Cut your losses early. If you have a gut feeling that things aren’t working, act on it
- The importance of self-belief: The more you value yourself and your time, the more likely others will