Man, I’m pretty sure in the future we’ll just be writing .md files and sharing them. The SWE role will mostly involve creating detailed, well-documented examples and LLMs will handle the actual coding.

Beware of premature optimization.

Watching those up and down arrows in Claude Code is so satisfying! Add that to the tokens going brrrr

Is it just me or has Gmail been extremely slow recently?

This is incredibly cool!

Incredibly impressive. The fact that the model was able to solve complex, unverifiable problems with enough accuracy to win gold at the IMO is a huge breakthrough. Congrats to the OpenAI team!

Since I started using Claude Code, I mostly open the IDE just to read code or make minor tweaks, especially for greenfield projects. With async agents becoming more ubiquitous, I can definitely see a world where this becomes the norm. Twelve months is a tall order, but the pace

In built router to switch between reasoning and non reasoning omg I am so pumped!!! Hopefully it’s not ridiculously priced.

NIK Sam Altman The continuous RL improvement of Grok feels like AGI. Grok 4 today is smarter than Grok 4 a few days ago.

When does it even make sense to use API pricing with Claude code? I feel like I’ve already burned through the credit I uploaded. If I don’t switch to account-based pricing soon, I’ll probably end up spending a lot more.

This is an incredible use case. I’ve been using an agent to write SEO friendly blogs for a side project and it’s been doing really well. Get X api to scrape trends -> write a blog about the topic -> manual email approval -> publish. Works like a charm!

Honestly, we are about to enter bot-apocalypse. All content in most places will be AI generated. User generated content will be the new gold.

Crazy week

Why is this an owl?

🚀🚀🚀

I feel Claude Code’s plan mode should be more aggressive when it comes to asking clarifying questions. Some of the arch decisions it takes on its own are somewhat questionable.

XML prompting is the real GOAT!

I really like Grok for sentiment analysis. It’s ridiculously helpful when it comes to evaluating what folks in the tech community think about a product or a platform.

Wikipedia is really underrated. It’s a great place to brush up on quick facts and, for the most part, has just the right amount of detail to get a solid understanding of a topic.

Open model!!