Speed Is THE Primary Business Strategy
Mike Cassidy on making speed a single, consistent strategy across fundraising, hiring, product development, sales, marketing, and pivots.
neversupervised
Latest posts
The Biggest Value of Math Isn’t Math
Learning hard things wires the brain for focus and persistence, and AI lets students skip the struggle that builds it.
Good Benchmarks
How to design realistic, verifiable, well-specified, and genuinely difficult agent benchmark tasks.
What makes a benchmark actually hard
Everybody is spending billions on hard benchmarks, and almost none of them are actually hard. The question of what makes a benchmark hard is one of the most important questions in evals, and almost nobody is working on it. Difficulty is ambiguous. It’s not understood. People keep commissioning harder tasks and they keep getting saturated faster than they come online.
The Verifier Is the Hard Problem
I’ve been contributing to Terminal Bench since close to the beginning. A lot of my interest lately has been with reward hacking, specifically how elusive it is. I thought it was an obvious, important issue, and when you talk to people they say, sure, reward hacking is a big deal. But it keeps getting bigger. I think it’s a bigger problem than most people thought.
Why I Invested in ChipAgents
I invest in companies where AI is going to change how things get done. There are businesses that look like businesses you could have founded five years ago, and some of them will still do great, but that’s not my area. My area is businesses that couldn’t have existed before.
Goals, Motivation, and Shipping
This is a transcript of a mentorship conversation I had with an AI Safety founder. It has only been lightly edited.
The Harness
I have rather high certainty that the way software is built is about to change dramatically, and that most vertical software will shift to general-purpose agents. The vertical SaaS economy is an economy of automating workflows that are the same every day; you look for the people who do the same thing fifty times a day, and you automate that. Software, in that world, is a workflow: a predetermined…
The Exhaust Is the Product
I recently spent some time at Constellation, exploring ways in which I can contribute to safety. Given my experience, I naturally look for commercial applications. I asked several people about their thoughts on whether commercial approaches to safety are viable. Nobody denied that an intersection probably exists, but I detected skepticism.
How Software Will Be Made
Lately, I only build things where I’m the sole author, and I 100% vibe code it. I don’t really look at code. I look at commits a little bit, but not really. I look at the app. I look at the behavior.