Clayton Kim
@TheClayMethod
CTO | Data Scientist | Painfully Mediocre Circus Artist. Building the future of human performance @fountbio. Former @BCG and @Wayfair
You might like
It’s the year to put a harness on the fire horse
the startup dream team has shifted from hipster, hacker, and hustler to auteur, autist, and aurafarmer
Just prompt it. Excessively complicated agentic workflow engineering is going to be eaten up by better models anyway
I'm afraid tokenmaxxing loopgooners will get totally framemogged by high T promptchads and visionstaceys that practice disciplined ideamewing
I built agent skill that implement @boristane 's workflow to research the codebase, write a plan, let the user annotate the plan until it's right, then execute without stopping. github.com/theclaymethod/…
Got bored one the plane and I codified my AI copy-editor designed to remove slop from AI writing into an open source Agent Skill claytonkim.com/writing/unslop
Ai designers should learn conventional design Conventional designers should learn Ai design
Old: just slap a gradient on it New: just use a shader… shader it all
I really don’t want to orchestrate agents all day to do the work that I actually enjoy doing, this is super miserable state of existence, I struggle to understand how my colleagues enjoy this
texting her: bye gn going to sleep ❤️ me and claude until 3 am:
Especially the part when it all goes wrong
Just shipped /last30days. A Claude Code skill for @claudeai that scans the last 30 days on Reddit, X, and the web for any topic and returns prompt patterns + new releases + workflows that work right now. Last 30 days of research. 30 seconds of work. 👉 github.com/mvanhorn/last3…
A Claude code skill that forces you to answer a short quiz about the code that you're about to commit: github.com/theclaymethod/…
Claude Code is humbling in how fast it can prove that my cool backlog ideas that I never had the time to implement were actually pretty mid
A little Claude code doodle: a chat UI that uses rapid serial visual presentation (RSVP) to stream tokens straight back into your brain. Most of reading is actually moving your eyeballs, which is pretty slow.
Sometimes senior people at work will complain that the task you gave wasn’t completely well defined with 100% of specifics. Homie you get paid to fill in the gaps. If I had an airtight spec I’d be giving it to an LLM. Don’t coach me into making your job an LLM job.
Blue zones basically deserve their own logical fallacy category
I’d hate to say it, but good code feels good. It’s like taste, but for abstraction
A few things I've noticed as all devs write code with AI. When you write foundational / architectural code of a new project by hand, you "feel" the code pushing back if your abstraction isn't right. You feel when something is harder than it should be. The code is telling you
United States Trends
- 1. #USAHockey N/A
- 2. Good Sunday N/A
- 3. Go USA N/A
- 4. Muhammad Qasim N/A
- 5. #sundayvibes N/A
- 6. USA vs Canada N/A
- 7. Greenland N/A
- 8. Kurt Russell N/A
- 9. Sidney Crosby N/A
- 10. #GoldMedalGame N/A
- 11. TSA PreCheck N/A
- 12. Blessed Sunday N/A
- 13. USA USA USA N/A
- 14. Mike Eruzione N/A
- 15. Miracle N/A
- 16. Seek His N/A
- 17. Go Team USA N/A
- 18. LETS GO BOYS N/A
- 19. Ryan N/A
- 20. USA ML N/A
Something went wrong.
Something went wrong.