Pulkit

3.8K posts

Pulkit banner
Pulkit

Pulkit

@puhlkit

Interested in education, fitness, technology, and finance.

United States Katılım Kasım 2024
200 Takip Edilen148 Takipçiler
Pulkit
Pulkit@puhlkit·
100% sure they updated GPT it is now so much nicer to talk with
English
0
0
0
11
Pulkit
Pulkit@puhlkit·
Sold 480C $MU 3/20
English
0
0
0
41
Pulkit
Pulkit@puhlkit·
@peligrietzer True i thought they had surveyed every single person on earth
English
0
0
9
1.8K
Pulkit
Pulkit@puhlkit·
I have finally started enjoying tweeting and I feel like I have figured “something” out
English
0
0
0
7
Pulkit
Pulkit@puhlkit·
@aakashgupta Never used copilot because I just assumed it’s as bad as teams
English
0
0
1
314
Aakash Gupta
Aakash Gupta@aakashgupta·
Nadella paid $650 million to acquihire Mustafa Suleyman and 70 Inflection employees in March 2024. The job: make Copilot the AI product that justifies Microsoft’s infrastructure bet. Two years later, Suleyman no longer runs Copilot. The corporate framing is generous. “Freed up to focus on superintelligence.” The numbers tell a different story. Microsoft 365 has 450 million paid commercial seats. After two years on the market, during the largest AI hype cycle in history, Copilot converted 15 million of them. That’s 3.3%. At $30/user/month, those seats generate roughly $5.4 billion annually. Microsoft spent $37.5 billion on AI infrastructure in a single quarter. The competitive data is worse. Recon Analytics surveyed 150,000+ enterprise users in January 2026. Copilot’s paid subscriber share dropped from 18.8% to 11.5% in six months. Gemini passed it in November. The most damning finding: 70% of users initially preferred Copilot because it was already embedded in their Office apps. After trying ChatGPT and Gemini, 8% kept choosing it. That 70-to-8 drop is the number that explains this entire reorg. Microsoft has the greatest distribution advantage in enterprise software history, and 90% of users leave after trying the competition. So Nadella hands Copilot to Jacob Andreou, a former Snap executive. You bring in an eight-year consumer growth operator when the problem is adoption, not science. And Suleyman gets “superintelligence”: no shipped product, no revenue target, no quarterly earnings call where an analyst asks about the 3.3%. The $650 million acquihire just became the most expensive research fellowship in tech history.
Aakash Gupta tweet media
Pedro Domingos@pmddomingos

The inevitable has happened: Copilot no longer reports to Mustafa Suleyman. theinformation.com/briefings/micr…

English
79
164
1.4K
382.8K
Take Testosterone
Take Testosterone@maxyourtest·
New libidomaxxing method just dropped
Take Testosterone tweet mediaTake Testosterone tweet media
English
190
330
9.3K
1.5M
Pulkit
Pulkit@puhlkit·
There’s a case to be made that the models do better work when they have less context
English
0
0
0
6
Stark Capital
Stark Capital@trades_hayes·
@puhlkit @aincomeinvestor So the first apartment I had it worked in 2024, but it was paid a weird name and rocker money called it a “home and garden” category. Now at a much nicer place it doesn’t work
English
1
0
0
360
BRICK BY BRICK
BRICK BY BRICK@aincomeinvestor·
Hear me out. Robinhood credit card perk that earns 1-2% back on rent and mortgages.
BRICK BY BRICK tweet media
English
33
5
322
80.1K
Stark Capital
Stark Capital@trades_hayes·
@aincomeinvestor They don’t let me pay rent with mine, they literally told me it doesn’t work for rent or mortgages
English
4
0
3
4.8K
Pulkit
Pulkit@puhlkit·
@Ben27015572 Many reasons. - robinhood is a better product/company than coinbase - robinhood will make a better UI - stock research is more credible than crypto research - robinhood will monetize it the right way
English
1
0
1
28
The stranger
The stranger@Ben27015572·
@puhlkit So why does Robin Hood think that they’ll be successful at it when Coinbase failed ?
English
1
0
0
22
Pulkit
Pulkit@puhlkit·
Nobody is prepared for the realization that all jobs are just reading/writing text
Jack Raines@Jack_Raines

One of the funnier things about the AGI/pilled folks is that people are extrapolating what happened with coding automation to the rest of the world, but I don’t think that extrapolation holds up. Coding is, at its core, the generation of texts and numbers to tell a computer to do things. There is, obviously, a lot more to it then that. Like understanding how systems connect / should flow together matters as you can better think through what the complete code “should” look like. But to build that system, the primitives are letters and numbers. LLMs are trained on a gazillion examples of letters and numbers. Much of that is code (thank you Stack Overflow). They just take text inputs and make text outputs. Now, there are derivatives to this. Like LLMs can now call tools, and the outputs of those tools can be used as inputs to trigger other things, but they’re just writing letters and numbers which a harness then knows to “trigger” to do a thing. Which means the things that can be fully automated are things where the input and output are purely numbers and letters. So, writing (marketing copy, emails, books?!, code, Excel functions, etc.) But most other domains just have more nuance or friction or edge cases that, even if a lot of the flows can be automated, have something that they system can’t “get” or “do” that throws off the whole loop. And basically any job that isn’t purely “read/write text” has a lot of those friction points. So SWEs kind of built a thing that replaced their jobs, but I just don’t see that pattern matching to too many other fields.

English
0
0
0
21
Pulkit
Pulkit@puhlkit·
Can someone at OpenAI refer me for a job that’s literally perfect for me
English
0
0
0
21