On X
Posts about software and AI: building products, programming and the AI tools I use.
159 posts
You're a first-year tech bro, right?
You just finished reading a Twitter thread by some guru who put together a demo in an afternoon. You watched a two-minute video where an agent opens the browser, searches for information, writes code, sends an email, and updates the CRM like…
Wait, does Claude's web app also use up the tokens in your plan?
WHAT. I hadn't used its web interface in years, and now I find out it's eating into my Fable usage.
OpenAI has a separate quota for its web interface. I thought Anthropic would do the same... in its defense, the web app has improved, but holy shit...
Two possible headlines for the new project
Which one do you like more? Tell me why
P.S. It’s a bit weak for now because I’m missing the trust signals
This month has been insane.
Every day I wake up to a new Codex reset. I think we’re already at 10+ in 2 weeks. And I’ve also had like 2/3 resets of Claude Code.
I feel really bad for Cursor users. To be fair, I did warn you.
A good “hack” for designing with Claude is to ask it to make 6 variations of a component for you.
Pick the one you like best and generate more variations.
Keep going until you get something you like.
You can use Opus 5 for this without having to burn through Fable
📊 I’ve built a portfolio analysis tool
I’ve spent a week cleaning up the code and improving the tool. Everything’s more polished, consistent, and nicer now.
From now on, all turn-based games should have a little chat connected to Codex/Claude so you can prompt while the enemy does its thing.
The speed of today’s models is incredible.
I felt like king of the world 6 months ago with any GPT Pro that took 40 minutes to answer you. Now the dumbest model gives you that answer in 1 minute.
These levels of intelligence at these speeds are unreal.
I’ve made a public version so you can play around with it.
You write some text. The tool marks each sentence’s length with colors.
It’s 100% free and private: your texts stay in your browser.
Have fun. It’s addictive. t.co/Vw7BoeixTi t.co/MYnxFCLboD
The world we live in is incredible.
Every now and then, I come up with a new prompt to throw at AI out of nowhere, and my humble reaction is, “Holy shit, why hadn’t I asked it this before?”
And then AI goes and answers me in a way I didn’t expect it to be capable of.
I’m clear that the bottleneck is my questions.
