https://nitter.cf/t.co/LYeQPBRdUu ; prev eng + growth @ https://nitter.cf/t.co/iETs26O8qa
San Francisco, CA
Joined October 2017
- Tweets47
- Following370
- Followers21
- Likes85
Americans' expectations of AI's impact on society, by income.
Middle-income Americans are the most pessimistic of the three groups, with 42% expecting a negative impact versus 16% expecting a positive one. Upper-income Americans are more optimistic, though negative expectations still outweigh positive ones.
Feng Zhang man.
Co-authored landmark optogenetics paper with two of the 2026 Medicine laureates, then helped pioneer CRISPR gene editing, which won the 2020 Chemistry Nobel.
All before age 32. Neither prize went to him.
Top 10 global companies by ad spend:
1. Alibaba Group
2. Amazon
3. PDD Holdings (Temu, Pinduoduo)
4. L'Oréal
5. Procter & Gamble
6. LVMH
7. JD[dot]com
8. Samsung Electronics
9. Google
10. Nestlé
Block Blast collected $127 million in ad revenue January–May 2026 (beating Candy Crush Saga). An on-device XGBoost model uses phone model, country and past ad-revenue data to tailor difficulty to each player’s advertising value and engineer winning streaks followed by unavoidable losses (“MultiClearBoostThenKillTrait”) when lucrative ads are available. Could vibe code the game in a few hours.
11 He has made everything beautiful in its time. He has also set eternity in their hearts, yet so that man can’t find out the work that God has done from the beginning even to the end.
12 I know that there is nothing better for them than to rejoice, and to do good as long as they live.
13 Also that every man should eat and drink, and enjoy good in all his labor, is the gift of God.
Ecclesiastes 3:11–13
Tommy Colitsas retweeted
r-1 led K1Bench, Verona’s human-adjudicated benchmark of production document parsers, with 85.20% of pages parsed without a material error—9.2 percentage points ahead of the next-highest system.
The benchmark used real-world K-1 packets supplied by a partner accounting firm. Reducto’s r-1 is built for exactly this kind of document, preserving both the content and the relationships that give it meaning.
One of the greatest heavyweights ever needed seven more seconds. Hendrickson never gave them to him.
LEGO (the toy) get's it's name from the Danish “leg godt” — “play well.” Named and founded by Ole Kirk Kristiansen in Denmark.
Sir James Dyson was obsessed with vacuum cleaners losing suction as their bags clogged with dust, to the point of creating 5,127 prototypes over five years (using cyclonic airflow, basically a miniaturized cyclone separator used by sawmills to spin sawdust out of the air).
Years of pushing through rejection by incumbents and clever use of a Japanese licensing deal to finance the company are equally important parts of Dyson's eventual success. But it all started with an obsession.
We do a lot of work in our professional and personal lives to find people that satiate some need. These workflows involve research (time spent collecting and reviewing context), coordination, and sometimes convincing/writing with craftsmanship.
Find me investment banking associates in NYC to network with. Find me a relator for beach houses in Cape May. Find me an expert in clinical trials that I’ll pay for a discussion. Find me a residential architect with a modern portfolio. Find me software engineers with 5+ years of experience in distributed systems. Find me hourly warehouse workers at Amazon in California from 2022–2025. Find me a single person 25-30 years old, presbyterian and over 6’2”. Find me partners at mid-market law firms based in the Northeast, over the age of 60. Find me venture-backed, seed-stage fintech founders with 1M+ in ARR. Find me adults with type 2 diabetes, A1c 7.5–10.5, on stable metformin, not currently using insulin.
These are all matching problems. One side wades through context to separate junk from something worth there time and the other tries to reach people where they provide value. Lots of song & dance and hours spent. Bad things that waste time get through and many good connections get missed. Good connections defined as discussions or relationships where there is a high chance of real value (note: value takes many forms, not always a deal struck) captured on both sides.
Can we handle this process with AI? I think yes, end-to-end (times show up on your calendar). A simple pane with agents pursuing people for each objective in our life, where the deliverable is scheduled time with profile-of-person X (or maybe, a probability that connecting with person Y will be valuable and one-click to meet them). Not a search engine, a results engine for connection. Even the more simple workflows listed above require research and coordination. The more complicated ones require deep context and evaluation.
Some caveats. There is nuance to each workflow, can this be captured in a horizontal product? I’d feel differently about a date vs a sales pitch showing up on my calendar (but what if both will be great conversations/connections!). This is close to the second point, focus is important. Should we pick one workflow where there is the most pain, or start broad to find that workflow within early technical and network constraints? Lastly, data collection here is hard. Can it be a one-sided marketplace to start? It becomes so much richer with onboarded context on both sides. Cut through the fluff to what a person has to offer and what there problems are with verified context, instead of us impersonating someone and going through traditional channels.
Re-read the list in the second paragraph. Do any of the use cases resonate? If this product worked, and the people you’re looking for showed up in your life, would you use it? From another perspective, can this be boiled down to context or is the art of selling/presenting yourself with a hinge profile, cold call, billboard, email copy, or search result here to stay?
