lmao zuck is out here gutting every mag 7 company with billion-dollar comp packages while Apple hr is offering a 10% annual bonus to convince these same engineers to stay

chinas on a winning streak man
dropped 3 models last week & open sourced all of them
they’re all as good as Claude, Grok & OpenAI at 90% of tasks
then dropped 2 video models that beat veo 3
unbelievable
i’m guessing they flip this trend at some point (once their models are ahead of americas) and close source but I’m impressed at the rate at which they’ve been able to keep up
truly thought open source models would be at least 2 years behind closed source
clearly i was wrong.

anyone else think these ai browsers are good but not great?
the ideal experience is basically “Claude code for browsing”
Vibe browsing
goal: fewer clicks & buttons pressed to reach the intended outcome
type a few words, watch the ai open up your browser, surf web, write emails, manage your calendar, write docs etc
it should intuitively know your next move without asking and do it for you
starts with the mundane tasks (reduce tabs, organise calendar)
but eventually is present in every online interaction you ever have

models are actually good enough to do cool shit for you, its just the lack of context thats stumping it
anyone can give a model access to a tool or database but it doesn't know when to utilize it
solution probably consists of some combination of short-term + long term memory which can help the model structure its thinking for each incoming prompt before it executes
good article on this here

its time to accept the fact that tech companies / platforms now hold as much leverage (if not more) than the government of the nation they reside in.
trump parading sama, jensen & sundar to the saudis for a $2T investment in US infrastructure is just another domino falling in an already accelerating trend since the facebook cambridge analytica saga
those with the highest form of leverage right now are in AI simply because the tech is a "10,000X power-up" to whatever it gets applied to
thats why zucks throwing out $1B salaries to poach researchers
thats why the US defense dept has signed $500M+

getting a lot of people asking me “when do you use ai”
ans: im defaulting to using it for everything and so should you
figuring out what to do at the gym? Gpt
need a post gym meal? gpt recipe
don’t understand a tweet, concept or research paper? Grok, GPT
browsing the internet? Perplexity Comet
automate some mundane workflow like scheduling meetings? OpenAI agent, Claude
the idea here isn’t to become some kind of ai soy-boy but to
*learn how to leverage what’s going to be the most important tool in your arsenal going forward*
it’s going to replace your job anyway, might as well get ahead of

we are literally watching crypto & finance blur into the same thing & everyones moaning about prices
treasury co’s have ~$5B put aside to buy ETH (so far) & more treasuries launching every week
BTC & ETH etf inflows have been uponly
X is integrating Polymarket prediction markets
US gov encouraging open source AI & bending the rules (they don’t want to lose to china)
Debt skyrocketing while btc breaks $120K
the signs are staring you in the face.

people who master the art of prompting are going to be very well-paid
but whoever creates the software that does this for you is going to be extremely well paid
unlocking infinite value of AI today can pretty much be boiled down to how well you prompt
no better investment of your time rn

one thing bitcoin did really well was teach us how to build a robust, defensible compute network
its laid a perfect blueprint for whats to come in AI, data centers will need to become distributed
Xai, Meta & openAI have *each* committed $500B+ to build data clusters over the next 5 years and thats just for training models at individual locations
if we want to build a defensible economy powered by AI i think distributed compute networks is not just an option, its necessary.

one thing i'm loving is founders challenging the law to push innovation forwards
kinda how uber operated illegally in states to prove (with real data) that there was demand and suitability for strangers to ride in other stranger's cars.
and now Polymarket bringing prediction markets (aka betting) to the American consumer.
if you want to create a 0-1 product you sometimes have to be willing to literally break the rules
the rule-makers catch up after.

we are living in insane times
people are getting offered $1.25B to work on AI at the same time Harvard grads are struggling to find jobs 6 months after graduating.
we’ve never been in such a polarising environment economically
you have unfathomable wealth concentrating in <200 people whilst prospective geniuses fall to the wayside.
the good news:
- this will force talent to evolve into AI experts (this is crucial for USA to maintain the lead)
- this will force non-AI talent to adapt and create brand new (and VERY lucrative) economic hot-zones that have never existed before (we’re starting to

sat on a park bench reflecting and ive got to say im extremely optimistic about the convergence of crypto & AI
both technologies are entering their prime & are absolutely crucial for the advancement of humanity
sure v1 was a hot mess, we predicted a lot of things incorrectly - but that doesn’t take away from the fact we’re, without doubt, hurtling towards a world where “super intelligence” & “24/7 finance” is integrated into EVERYTHING we touch.
you’re a kid starting school? You’ll have a personalised AI teacher.
you’re scrolling a social media feed? Prediction markets tell you what’s real a