we’re thinking of killing plan mode and using the shift+tab hotkey to adjust effort levels
I don’t think the models need plan mode anymore, but if you’re a plan mode diehard would love to get your feedback on why
Looks like $META is building Meta Pay for Muse, a first-party wallet that would let its AI agent use saved payment methods to make purchases for users.
Genuinely cannot believe how well the Chinese are doing at training models given how absurdly poor they are relatively to the Americans. Tighter constraints make you write better software. Throwing money at the problem is a seductive yet ultimately destructive solution
Genuinely cannot believe how well the Chinese are doing at training models given how absurdly poor they are relatively to the Americans. Tighter constraints make you write better software. Throwing money at the problem is a seductive yet ultimately destructive solution
The real answer is they don't "own" the consumer anyway:
Booking spends ~7B a year on Google Ads
Expedia spends ~6B a year on Google Ads
They spend a fair amount on Meta as well but nowhere close
If anything, this partnership gives them some line of sight to reduce dependency on Google
Funnily the inverse is true for many consumer brands
Ads is a much bigger industry than anyone expects, more people spend on ads a year than all of SaaS put together many times over
Openai and Anthropic have paused model intelligence and are focusing on cost.
They don’t know how to stop the distillation of their models and then getting undercut from open models.
Of course they’re selling this as ‘pacing the frontier’. But this is a real problem that they’re scrambling internally to solve
"Waymo is disrupting $UBER"
Waymo started offering paid AV rides directly through their own app in mid-2024
By the end of 2026, they are projected to have completed around 50-60 million total paid rides (over 2.5 years)
In that same time frame, Uber will have completed around 35-36 BILLION trips (around 18-19 billion mobility)
That means Uber will have done around 336x as many mobility rides as Waymo...
In 2026 alone,
- Uber: 16-16.5 billion trips (~9 billion mobility rides)
- Waymo: ~30-40 million mobility rides
Or 257x as many mobility rides (Uber will do around 25 million mobility rides PER DAY)
And this is at a time when Waymo is the ONLY AV with any type of scale in the USA
What happens when Lucid, Wayve, Zoox, Rivian, Avride, Motional, May Mobility, etc join Uber's platform? And scale?
LLMs have plateaued since opus 4.8 and OpenAI & Anthropic are running on thinner margins than ever to ensure their models can justify the 10x premium over alternative models
JUST IN: Startups are increasingly building custom AI models using open-weight alternatives to cut costs & reduce dependence on OpenAI & Anthropic. — Bloomberg
seeing takes where people are downplaying $AMZN for not being forward thinking on agentic commerce
here’s a quote from Warren Buffett:
“Lose money for the firm and I will be understanding; lose a shred of reputation for the firm and I will be ruthless.”
$AMZN could open up this new shiny agentic distribution for a quick buck, at the risk of losing its reputation as the best marketplace experience
why would they? i’m sure a middle ground will be reached where $AMZN, $META, agent users and sellers can all benefit from this agentic revolution — just a matter of time
Amazon cuts off Muse.
While I am bullish Meta and Muse, I think many people are overlooking the digital knife fight that’s about to occur
Nobody wants to get commoditized or layered here. Let the games begin
hyperscalers’ capex will contribute directly to $INTC $AMD the next few years behind the transition from inference to agentic capabilities with strong demand of CPUs