Ash Goyal

FSD recognizes officers hand signals. FSD 14.2.2.5 stopped for a police officer who was assisting in a fallen branch and directing traffic. FSD waited until the officer waved me on to go, the car recognized his hand signal.

Fireside chat at Sequoia Ascent 2026 from a ~week ago. Some highlights: The first theme I tried to push on is that LLMs are about a lot more than just speeding up what existed before (e.g. coding). Three examples of new horizons: 1. menugen: an app that can be fully engulfed by LLMs, with no classical code needed: input an image, output an image and an LLM can natively do the thing. 2. install .md skills instead of install .sh scripts. Why create a complex Software 1.0 bash script for e.g. installing a piece of software if you can write the installation out in words and say "just show this to your LLM". The LLM is an advanced interpreter of English and can intelligently target installation to your setup, debug everything inline, etc. 3. LLM knowledge bases as an example of something that was *impossible* with classical code because it's computation over unstructured data (knowledge) from arbitrary sources and in arbitrary formats, including simply text articles etc. I pushed on these because in every new paradigm change, the obvious things are always in the realm of speeding up or somehow improving what existed, but here we have examples of functionality that either suddenly perhaps shouldn't even exist (1,2), or was fundamentally not possible before (3). The second (ongoing) theme is trying to explain the pattern of jaggedness in LLMs. How it can be true that a single artifact will simultaneously 1) coherently refactor a 100,000-line code base *and* 2) tell you to walk to the car wash to wash your car. I previously wrote about the source of this as having to do with verifiability of a domain, here I expand on this as having to also do with economics because revenue/TAM dictates what the frontier labs choose to package into training data distributions during RL. You're either in the data distribution (on the rails of the RL circuits) and flying or you're off-roading in the jungle with a machete, in relative terms. Still not 100% satisfied with this, but it's an ongoing struggle to build an accurate model of LLM capabilities if you wish to practically take advantage of their power while avoiding their pitfalls, which brings me to... Last theme is the agent-native economy. The decomposition of products and services into sensors, actuators and logic (split up across all of 1.0/2.0/3.0 computing paradigms), how we can make information maximally legible to LLMs, some words on the quickly emerging agentic engineering and its skill set, related hiring practices, etc., possibly even hints/dreams of fully neural computing handling the vast majority of computation with some help from (classical) CPU coprocessors.

It’s like it can see the future sometimes

Only 1 of the top 10 countries of origin for migrants has a lower median age than those born in Australia. That country, Nepal, is 8th on the list of countries of origin by population and is dwarfed by the top few. So the overwhelming majority of first generation migrants are older than people born in Australia. Which means that our population isn’t getting younger as a result of migration. New migrants are mostly younger - but to actually reduce our population age requires more and more younger and younger migrants. Which is contributing to a lower and lower local birth rate as the adverse economic impacts of migration deter Australian-born children. This is yet another in a long series of black pills for me.

He worked in a lab so small & hot that his sweat would often ruin his notes. He did not build a bomb/a satellite, but he solved a puzzle that was killing millions. In 1959, from a Silo in Calcutta, Sambhunath De discovered the secret pump that drains human life. He is the father of ORS: the ghost behind the most successful medical intervention in history. He was nominated for the Nobel by the world’s greatest giants, yet he died in 1985 traveling by local bus, unrecognized by the very people whose children he had saved. Born in 1915 in a small village in West Bengal, S.N. De did not come from a family of elite scientists. He worked at the Nil Ratan Sircar (NRS) Medical College, Calcutta. While elite scientists were building rockets, De was working in a tiny, cramped lab with barely any ventilation. He did not have high-end sensors. He used his own intuition & rudimentary tools to study how certain invisible forces acted on human cells. S.N. De solved the mystery of Cholera, but in a way that was pure Fluid Physics & Biophysics. For a century, the world thought Cholera was a blood infection. In 1959, in his tiny lab, De proved it was a toxin that attacked the fluid-transport mechanisms of the gut. He discovered the Cholera Toxin (CT). He demonstrated how the toxin created a pump that sucked water out of cells, a masterclass in osmotic pressure & molecular transport. This discovery is the direct reason why ORS (Oral Rehydration Salt) exists. If we/anyone we know has ever been saved by a packet of ORS, we owe our life to S.N. De. In 1954, Nobel Laureate Joshua Lederberg wrote to the Nobel Committee saying that S.N. De’s work was worthy of the prize. He was nominated multiple times, but like many Indian scientists, he was a Ghost in a colony. The prize never came. In 1978, the Nobel Foundation organized a symposium on Cholera. They realized the man who started it all, S.N. De, was still alive in Calcutta. They invited him, & he arrived at the high-end gala in a simple suit, looking like a retired clerk. He was a man of aggressive humility. He lived in a small house, traveled by local buses, & never sought patents for his discovery. He wanted the Signal (the cure) to be free for the world. His own family knew him as a dedicated doc who went to the lab every day. They had no idea that Nobel Prize winners in Europe & America were referencing his 1959 paper as 1 of the most important scientific documents of the 20th century.

BREAKING: Massive fire erupts at HPCL's ₹80,000 crore Pachpadra Refinery in Balotra, Rajasthan — just one day before PM Narendra Modi's scheduled inauguration. The 9 MMTPA greenfield refinery-cum-petrochemical complex, dubbed the "Gem of the Desert," was set for commercial operations in July. Situation under control, probe underway. Multiple agencies likely to be roped in!