"Moat" is not something that's logically deduced; it's something that's built. "TAM" is not something that's calculated; it's something that's built. Execution is the only moat.

Great post on unconventional distribution tactics

Among the media formats (text, audio, and video), audio seems the most ripe for disruption rn. 1. For text - ppl don't read 2. For video - the tech is not there yet; AI-generated videos still feel "too AI"

You can’t outsource taste

I really like the term “context engineering” over prompt engineering. It describes the core skill better: the art of providing all the context for the task to be plausibly solvable by the LLM.

Is there a “Perplexity for Notetaking” yet? I realize I take a lot of Apple Notes but per Apple being not very good at software anymore, the chance of it becoming smarter anytime in the next 24 months is zero Eg if I paste a link I want the important or interesting parts inline

Great piece. Despite advances in machine translation, the number of available job opportunities for translators and interpreters has actually been increasing. ingrids.space/posts/what-goo…

My favorite hotel is one in Shanghai where there was literally a gym in my room

There’s no “overnight success”, even in AI. Most successful AI startups have been around for far longer than we think. Most AI coding startups were founded before ChatGPT launched. When OpenAI launched GPT3.5, they’d already been around for 7 years.

How to overcome "AI FOMO"? Find your place. This is how I think about my positioning in the AI wave:

Hit the In-N-Out drive-thru with Roy, 21-year-old CEO of Cluely - rescinded from Harvard, expelled from Columbia, then raised $15M from a16z in a week after bootstrapping to $3M ARR profitably in 60 days by grabbing hundreds of millions of eyeballs through his

"You have to ask: why did personal computer get started in Northern California in the late 60s when at the time every major tech company in the country was on the East Coast? And the answer is: because there was a very small group of people in and around Stanford that saw

A lot of ppl ask “How do I build an audience online?” You’d think that they’d already posted a lot and were failing to attract attention. But no. 99% of the time, they haven’t even posted anything. Half of winning is just “starting”.

There are way too many AI products to try and way too few problems to be solved by them Normal people just don’t have that many use cases in their lives

“Luck” is unevenly distributed. One can increase one’s “luck surface area” by: - doing crazy good work - having a great reputation - building genuine connections - building an audience/a discoverable body of work

This is the only tech podcast as fun as a comedy show. Love it Casey Newton Kevin Roose youtube.com/watch?v=cT63mv…

So many AI products want to help me create stuff from scratch, but what I really need is for AI to polish stuff I already have. The former is for the amateur/hobbyist market. The latter is for serious work by professionals.

We need more “humble AI” that invite humans to contribute. Too many AI products have hubris and try to do too much. Kudos to Chris Pedregal Granola for creating a “humble” product More: zarazhang.substack.com/p/we-need-more…

Building Netflix taught me that your first idea is rarely your best idea. We started mailing DVDs, not streaming movies. Stay married to the problem you’re solving, not the solution you THINK will solve it.