The fastest bookmark experience ever made for X.
New York, NY
Joined July 2026
- Tweets1.6K
- Following836
- Followers656
- Likes806
Big day: ZeroMark is now a Claude connector 🎉
claude.ai/directory/zeromark
Your X bookmarks, finally useful.
@claudeai now reads your taste in X bookmarks: every post you kept, every one you let go, every note you wrote.
before any app remembers things for you, ask:
where is it stored
can i export it
can i delete it all in one click
does it train on it
what happens if they shut down
if they can't answer in one line, don't hand them your brain
cleared 1 bookmark today
one person clapped
that's infinite growth and i will be taking no questions #zeromarkit
3 questions i ask before i hit save
1. will i use this by friday
2. who would i send it to
3. did i already save something like it
no to all 3? like it and keep scrolling
saving is a promise. a like is just a nod
Before building an agent for architects, I would ask an architect to show me a plausible drawing they would reject.
Then ask why.
Those explanations could become the foundation of the product's evals. The interesting knowledge lives in the reasons an expert says no.
new campaign promise:
end "read later"
nobody has ever read it later. i checked.
small candidate, big promises #zeromarkit
reply with a screenshot of your oldest x bookmark
no context. no explanation
we'll guess why you saved it
50 unpublished physics problems, each about a day's work for a world expert
best ai agent solved 49 of them, under $3 each
meanwhile the physics thread in my bookmarks is still at 0 of 1 xitter.cf/rohanpaul_ai/status/21…
New paper from Google Deep Mind, and many top Universities finds the best AI agent tested solved 49 of 50 unpublished research-level theoretical physics problems, each roughly a day's work for a world expert.
at under $3 each.
Benchmarks for AI in physics often use textbook or contest problems, or block the tools real researchers use. So it was unclear how agents handle the calculations that fill daily research.
Experts contributed 50 unpublished problems from their own research, each about a day's work for a specialist. Agents worked alone, with and without web search, and were graded only on final numerical answers.
GPT-6-Astra in Codex solved 49 of 50 with web search, and both it and Opus 5.5 in Claude Code solved 46 of 50 without it. Accuracy barely changed without search, which suggests the models have memorized much of the physics literature.
i am hiring!!!!
need someone to read my 2,000 bookmarks for me
pay is $0. benefits: you will know literally everything
requirements: must never say "i'll read it later" #zeromarkit
x is adding long press on like to send someone $1
a like might finally mean something
if bookmarks cost a dollar each, i'd be broke and i still wouldn't have read them