Recent Notes
View all> It doesn’t matter how fast you move if it’s in a worthless direction. Picking the right thing to work on is the most…
> You have to both pick the right problem and do the work. There aren’t many shortcuts. If you’re going to do somethin…
> three key pillars: “Make sure to get the important shit done”, “Don’t waste time on stupid shit”, and “make a lot of l…
"I think it’s good to overcommit a little bit. I find that I generally get done what I take on, and if I have a little…
generally try to avoid people and situations that put me in bad moods, which is good advice whether you care about produ…
**Positioning and packaging drive a product's success** - Positioning is critical and powerful, even if Clear says he h…
**Reading and writing both matter** - Reading is filling the car up with gas. Writing is driving and going somewhere. Yo…
Getting more output from each unit of work: leverage - Leverage is getting more output out of each unit of effort. Some…
Work on things with a long half-life - A long half-life means the work keeps paying off long after you make it. Clear d…
One interesting question to play with is what do you do when there's nothing to do? A lot of people when they are standi…
It is so unlikely that whatever you are working on now is the best use of your time — it's almost impossible. And so I t…
## "having fun" makes you hard to beat - For any habit you want to build, ask "What would this look like if it was fun?…
"Consider, for example, the famous “Toyota Production System,” the philosophy that determines how Toyota makes its cars.…
## Supermodularity ### Source - Comes from a paper by Paul Milgrom and John Roberts, called “The Economics of Modern Ma…
"The last week was funny and also tiring, I worked 14 hours per day on average. My normal average is 4/6 since early Red…
If I had to ask myself the question — had I the right to gain power? — then I certainly hadn’t the right
### The two-minute rule - Take whatever habit you're trying to build and scale it down to something that takes 2 minut…
- Every action you take is casting a vote towards being the type of person you want to be - Act of doing it is proof you…
Change in the environment such that the behaviour that you want is obvious. You want to be able to walk into your rooms…
In a lot of ways, I actually think the most powerful form of mental toughness, the most powerful or resilient form of pr…
June 2026
May 2026
Vercel achieved an 18x speedup by accepting a harder constraint—treating every build as potentially malicious—and building from scratch with Firecracker microVMs instead of containers, then stacking three optimizations on top.
Vercel achieved an 18x speedup by deliberately choosing a harder foundation—microVMs instead of containers—because their threat model (running untrusted customer code on shared hardware) demanded it, then stacking three optimizations on top.
Vercel achieved an 18x speedup by accepting a harder constraint—treating every customer build as potentially malicious—then building from scratch with Firecracker microVMs instead of containers.
The 100x productivity gap in AI coding isn't about smarter models—it's about architecture: push intelligence into reusable markdown procedures (fat skills), execution into deterministic tools, and keep the orchestration layer (harness) razor-thin.
Toyota workers stop the production line 2,000 times weekly to fix problems; Ford workers do it twice. That single metric reveals why GM's attempt to copy lean manufacturing is closing the productivity gap but missing the cultural transformation that actually matters.
Japanese firms like Toto (toilets) and Yamaha (pianos and motorcycles) aren't just diversified—they're a fundamentally different organizational species optimized for incremental refinement over shareholder returns, explaining both their dominance in precision manufacturing and their absence in software.
AI video editors assume your footage is already labeled—but the real bottleneck is making unlabeled archives queryable in plain English, which you can solve locally for $22/month instead of $140 in SaaS rents.
Google's new Gemini 3.5 Flash costs 3-6x more than previous Flash models, yet they're deploying it to billions of free users—a bold bet that API customers will absorb price increases while consumer deployment drives ecosystem lock-in.
A top-tier CTF competitor explains how frontier AI models (GPT-5.5, Claude Opus 4.5) have turned security competitions from skill ladders into pay-to-win orchestration races, breaking the primary pipeline for developing elite security talent.
GGUF files contain chat templates, sampler configs, and special tokens—but missing metadata for tool calling formats, think tokens, and multimodal projections still forces developers to write model-specific code.
Why building small, "unimportant" software projects for personal curiosity is more valuable than the tech industry wants you to believe—from the creator of Redis on his post-retirement work.
Your codebase is a Marble Madness level and your AI agent is the marble—your job is removing hazards so it rolls smoothly, plus a vision of "Barbapapa software" that reshapes itself while running to fit your exact needs.
A simple two-step technique for generating AI images with accurate text and numbers: create a precise SVG "underdrawing" with correct layout, then have the AI model paint over it to make it beautiful.
James Clear reveals why ambitious people fail by optimizing the wrong variables - and his exact systems for life positioning, from deleting all social media to sequencing advantages across decades.
DeepSeek just released the largest open weights model ever (1.6T parameters) that performs within 3-6 months of GPT-5.4 and Gemini-3.1-Pro but costs 10-20x less, with efficiency breakthroughs that use only 10-27% of the compute for long context windows.
AI is about to trigger the biggest SaaS churn wave yet—not in 2024 or 2025, but in 2026 when contracts renew and companies finally have time to rebuild their stacks around agents, internal AI tools, and new workflows.
AI isn't just changing what software does—it's breaking the SaaS business model by enabling internal builds, forcing vendor consolidation, and making 18+ month CAC payback periods mathematically impossible when customer retention collapses.
When you hit the afternoon focus wall, the solution isn't more coffee or a break—it's literally staring at a wall for 5-10 minutes to reset your dopamine-fried brain.
Cursor trained their AI coding agent to learn what information matters by making self-summarization part of the RL training loop, enabling it to solve problems requiring hundreds of actions that exceed its context window.