Surya Dantuluri

I grew up in the Bay Area fine-tuning GPT-2 to do my homework.

I previously built a ChatGPT plugin with 1M+ users, co-founded a defense co backed by OpenAI, and worked on computer-use models.

I built models for long-tail MEV with GPT-3 during college and for prediction markets with o1 deep research, ranking #1 on Polymarket by monthly PnL.

I now work at an AGI lab diffusing AI across the economy.

Writing

Projects ↗

AGI diffusion

We have digital intelligence, but it's so unevenly distributed that much of the world still operates like it's 1990.

I can do almost anything from my phone, but people still drive to offices that haven't changed in half a century. We have models that code better than most engineers, yet drug discovery still takes decades, total factor productivity has stalled since the 1970s, and our power grid runs on software older than me. Somewhere along the way, we ended up serving the software that was supposed to serve us. Everyone talks about AGI while sitting at the same desks, staring at the same screens, solving rehashed versions of the same problems. Where's my flying car?

In each successive year since 2021 I found the economic output per token generated increases by an order of magnitude with each new modality—seq2seq, GPT-2, ChatGPT, deep research. Yet we keep using AGI to improve software instead of replacing software entirely.

AGI diffusion is the bottleneck to what we actually want: intelligence that works beyond the screen, diffused across the real economy where the problems that matter live. This may require significant capital and model training, but it certainly will happen with far fewer people who deliver insanely outsized impact. Once done, the question will not be whether an abundant future arrives, but why it was delayed at all.

Information-bound society: I believe AGI shifts us from a labor-bound to an information-bound society. It gives us compute-bounded leverage to act, producing exponentially more information until public knowledge becomes worthless and private information priceless. Capital, talent, and power flow to information asymmetry.
Information asymmetry: The biggest arbitrages emerge when you know something most relevant players do not. In academia, the process of becoming "great" is built on publishing your best work openly so you or I can work on top of it, though times are changing. Credentials and experience can sometimes be a way of justifying years spent within an institution. If you can learn anything in 2 weeks you can act, uncover new information, gather more information, and be part of the 0.1% who run the world. Break down walls, seek the truth, and don't be weird.
Focus: "Focus is saying no to something that with every bone in your body you think is a phenomenal idea, you wake up thinking about it, but you say no to it because you are focusing on something else." — Jony Ive
Blinders: "When you are a racehorse, the reason they put blinders on these things is because if you look at the horse on the left or the horse on the right, you're going to miss a step. When you're running after something you should not look left and right. What does this person think? What does that person think? No. Go!" — Jimmy Iovine
Trust Economy: Every introduction is a trade in trust. Expertise is everywhere; trust isn't. Short-term trust compounds financially; long-term trust compounds relationally. Know exactly what you value, and make it clear to others what game you're playing.
Against mediocrity: Most people, parties, etc. fade on an 80 year time horizon, let alone in 8 months. Put a date filter and go to 2009 or 2015, on WSJ or YouTube and you'll find a sea of passion projects started by founders who defaulted to fate over forging their destiny. Developing a sense of what mediocrity looks like helps you hire, fire, and promote ruthlessly. Stay away from people who are mediocre at all cost.
Up the intensity: You are a product of exploring every opportunity times your ability times the current state of the market. You can crank up your clock speed to the point where you are exploiting "life" and not exploring "opportunities". Some of the worst people I have worked with have no discernible clock speed or seriousness to the opportunity cost in which you or I could be working elsewhere. You can get a sense of seriousness if you're thinking about where else you can find a flow state in, instead of not working at all. Working somewhere for the status often times backfires because serious people will backchannel, far worse for you than fully pushing your weight elsewhere.
High Agency: Do free work everyday for years, which believe it or not, ends up compounding. Claude Monet painted for 15 years without making money, averaging a painting every 10 days for 60 years. Act on favors, curate, simply and genuinely care about others. People are always watching, many "connectors", "vcs" will also come along by because you have some alpha/tacit knowledge, remember never to give that away for validation.
Legacy as narrative: Every empire, religion, and revolution began as a story that refused to die. Great people, nations, and products shape narratives others want to share. Rockefeller won't be remembered for Standard Oil or Venrock, but for the narrative that reshaped American industry and etched him permanently into our collective consciousness.
Love for the Game: Replace ego with love for the game. Your work should be a labor of love, amplified with people who share this passion instead of chasing status. Status, like money, is fleeting if pursued directly rather than a byproduct of mastery with people who succeed together.
Speed Kills: Action—even without perfect information—creates momentum, which generates confidence, attracts talent, and then reveals signal. Speed itself is relevant, but often puts out a 'Bat-Signal' that compounds with secrecy.
Just Ask: If you don't know something, ask directly. You would be surprised how available 99.99% people are and the 0.01% are available if you have some agency. It's also much cheaper and faster to pay consultants to get this information.
Perceived Age: Each year can feel shorter than the last due to routine, dopamine fatigue, and lifestyle creep. The best way to fight this is to counterintuitively exploit every chance at everything you get, fully restart your life as many times as you can, and spend a few years doing nothing to understand your destiny.
AGI is here: It took over 200 years for people to realize Rome had fallen. It's clear to anyone talking to undergrads at most colleges that Artificial General Intelligence is more intelligent than the general population. Google didn't kill institutions; it transformed them. AGI will not kill businesses; it will reshape how business is done.