Dan Leshem

Does anyone know how I can get access to Antropic 100K LLM?

Chatting with the framework's technical documentation is developer experience on a different level

Today's wordle

When doing software 3.0 we can't assume Software 2.0 principles will apply

Turns out AI will replace developers and lawyers long before it will replace drivers

Headache. If I’d use that before I go to bed I had trouble sleeping. And overall, it was nice as an one time experience, but just too intense.

I sold my Oculus a month after I got it

Alternative moats: hard work. Serving your customers, offering UX/UI, creative marketing, out-of-the-box thinking, and a huge market.

-- What’s your tech moat? Investors often ask. Google's leaked memo proves nothing but the myth of “technological defensibility”. Investors should grow from this question.

How did I miss that ai.com links to ChatGPT?

This is just insane how much of my code is now generated by AI

AI agents will become an unfair advantage for many small startups.

In the near future, we'll see founders overseeing AI agents and achieving what a 500-people organization can't.

LLMs make side projects fun 🤩

Lovely. Austin is so beautiful ✨

Let’s see what happens w/ Bluesky

I’m looking to improve my note-taking system. What are some best note-taking workflows out there?

There’s an AI for that > There’s an app for that 🦾

ChatGPT folks, what are some good prompting guidelines for software development?

OpenAI “building in public” approach is fascinating. I won’t be surprised if this strategy is one of their top revenue drivers

How many of the BlueSky team are Ex-Twitter?

Forget AI, we now have Time
Machines!

So far it’s nice here

I wonder if, now that AI is taking over knowledge workers, traditional businesses offering physical goods and services will become more valuable.

Illustration: AI is eating the world

AutoGPT now has 109K stars on GitHub (a one-month-old project). For comparison, React has 209 thousand stars (a decade-old project that basically runs the internet). Wild.

Yo 👋

Researchers have reported success in increasing LLM's context length up to 2M tokens. For instance, GPT-4 model can take input size up to 32K tokens.

This might be addressing one of the biggest limitations of today's LLMs.

https://arxiv.org/abs/2304.11062

Hello World! 👋