One of the NBA's most interesting teams has completed 10 games — too early to make definitive conclusions but enough time for first impressions. What should we make of the Washington Wizards' performance so far? Story in @TheAthleticNBA.
https://t.co/ihyoQoYqsq
Democrats will now control the House and Senate and be able to hold Trump accountable. That's great.
At a time of massive income and wealth inequality, we must do more. We must aggressively fight for working families.
No more excuses.
1/ A few thoughts on tonight’s election:
Voters made clear they demand a different kind of country. Americans want an economy that can compete globally and deliver for them at home.
And they made clear that they want leaders they can trust to fight for their constituents.
Thrilled to be joining the board of @joinpursuit.
As AI changes how we all live and work, it's become clear to me that New York, and the country, needs a positive narrative about what AI makes possible. Pursuit is at the forefront of that story.
Pursuit teaches low-income New Yorkers to build with AI, and the results are life-changing. I've been inspired by builders who've gone from jobs paying around $40K to tech roles paying $100K+. A job like that changes things for a whole family, and for the communities around them.
If you're interested in getting involved (either in supporting the organization or in hiring builders) - shoot me a DM!
https://t.co/s8oMXowMpy
Anthropic will make ~1,000 employees worth +$100m at $2 trillion market cap.
There's ~11,000 Americans with $100m in investable assets.
So something like 1 in 12 Americans with $100m (liquid) will be from Anthropic. If you add in Openai, maybe 1 in 7.
Huge news from portfolio company Collective (@CollectiveFin).
They just raised $100M to build the AI financial OS for solopreneurs / Businesses-of-One.
Growing fast (ARR +120%+) and already helped members save ~$350M in taxes, and this month they’re tracking to add 70%+ of last quarter’s ARR growth.
Hiring a lot. The solopreneur boom is real.
Collective has raised $100M to build the future of solo business.
We’re building the AI-native financial OS that gives every Business-of-One the back office of a much bigger company.
Thank you to our members, our team and our investors.
Read the full announcement: https://t.co/9XwScZZied
As well as our CEO Hooman Radfar's blog post: https://t.co/jdbr7tBu63
Game 1 of 82 tips off @CapitalOneArena tonight.
I believe in our plan. I believe in the talent we are bringing together. Our fans are excited to see that work translate onto the court.
See you there tonight. Go Wizards!
4/ If your product doesn’t change to take advantage of longer runtimes, someone who does will be better at the same job and disrupt you.
Builders should be asking now: as agent length grows, what new value becomes possible, and what architecture and product choices does that require? You have to design for the next model iteration, not the current one.
The companies that treat longer agents as a product feature (rather than just a longer task) will own the next iteration.
1/The moat just moved again.
Until recently, AI agents worked in short bursts. They finished a task and started the next one from scratch. As reliability improves, runtimes are stretching from hours → days → weeks → months.
That changes what “a great product” even means in a category.
3/ When agents ran reliably for only a few hours, the best product was the one that completed a good search. Now the best product is the one that compounds context over time and surfaces candidates uniquely suited to that company.
Clients notice the difference. An agent that starts fresh every time will lose to one that has been learning for weeks.
This isn’t specific to recruiting. The same shift is coming to every category.
26,673,796 products identified.
A bin is a location. A product is a thing sitting in it, and one bin holds two or three, so products always outrun bins. Finding the location is a navigation problem. Working out what is in it is a perception one, and they are solved separately.
Every one of these was read off a label at height, from a moving platform, in whatever light the building had. Each read is also a training example, which is how 99.9% accuracy gets held rather than claimed.
The first million in our took seven months. The most recent million took eleven days.
See what that looks like in your racking: https://t.co/K8NKh1re6E
#PhysicalAI #ComputerVision #Logistics