Part of what makes agentic coding work today is knowing the mistakes. If you steer it back to an earlier state, you want the tool to remember what went wrong. There's value in failure. Don't let machines make the same mistakes.
Quoth the raven: 67.
I asked my nephew, a massive Ravens fan, what their record was a few weeks ago. He hesitated. He could rattle off every Ravens stat you could possibly imagine. But simply stating his team's record threatened to cheapen his fandom to a trivial meme.
I'm reading a physical book and I'm dreading copying my highlights into my digital notes.
I envision an app I can speak my notes into, and it shows similarly related ideas I've visited before.
I'm burning more tokens by speaking my prompts than typing.
Not because my prompts are longer, but because I write better than I speak.
When I write, I can think and edit my ask. Speech is rough and messy.
My Indian friend was complaining about the lack of diversity in his kids' public school in NJ.
"Not enough brown kids?" I asked.
"No. It's all Indians."
Some people are helpless when it comes to tech.
I used to wonder what it feels like to be so out of touch with technology.
---
My youngest was explaining brain rot words to me. I didn't understand a thing.
I get that feeling now.
Bad design makes money and itโs that money, in part, that pays our salaries. Thatโs part of why they pay us: itโs compensation for frustration.
โ @berkun
I had a shocking dream. At the end of it, something felt off. So I replayed the dream again in my sleep.
It was AI generated.
But I wasn't watching an AI video in my dream. My brain decided that my dream reality itself was generated. Weird.
I built a calorie tracking app.
I talk about what I ate, the Web Speech API transcribes it and gets sent over to OpenAI which calculates the calories and macros and logs it in Firestore.
I'm kind of blown away by how fast and how well this works even as a prototype.