–
- Developments in Generative AI a. Gen AI is becoming cheaper and more widely available for predicting the next token in a sequence. b. Gen AI is shifting from fast pre-trained responses to slower, more thoughtful reasoning during inference > this part specifically. c. Giving AI more computing at inference time (“test-time”) leads to better reasoning. d. Aside from ChatGPT, most generative AI models haven’t become successful application products yet. e. AI application companies are more than just simple user interfaces over a foundation model; they are not just simple wrappers but “cognitive architectures” (1), and these companies use multiple models, routing mechanisms, vector/graph databases, and compliance guardrails to simulate human-like reasoning. f. Traditional SaaS is evolving into AI-driven software that sells workflows and productivity tools.
- A big part of writing is to clarify and externalise your thoughts. (Sam Altman)
- Growth in companies comes from various innovations: products that meet specific needs, sustaining innovations that improve existing offerings, efficiency innovations that reduce costs, and disruptive innovations that create new markets. While each drives growth, disruptive innovations often have the most immense impact by reshaping industries and attracting new customers. (Where does growth come from?)
- The purpose of a startup is to grow, and retention is the most critical thing in that growth. (Alex Schultz)
- The root of great work is a sort of ambitious curiosity, and you can’t manufacture that. (When to do what you love)
- To create suspense, capture attention while slowly withholding key details, revealing hints. Lead readers toward one conclusion, then throw in a twist, keeping them guessing until the final, impactful ending. (Teller)
- Taste in science is critical for distinguishing between a good problem and one that no one will know the answer to—that’s taste. (Jim Simons)
- How to live an asymmetric life (huge upside): Don’t play to lose, do hard things, do what you love, do it for decades and write your story.
- Most of us spend too much time on what is urgent and not enough time on what is important. Success—in life and professionally—is often determined by making the time for meaningful, non-urgent tasks. (Advice from some Google engineer)
- Relationships don’t fail because there is some single catastrophic event to destroy them. However, often, there is a single catastrophic event around the end of the relationship, but that’s typically a symptom rather than a cause. What usually happens is it accumulates in the little things that just build up over time. (Scar Tissues)