@tszzl
@tszzl
you're young. When your joints start to hurt, you won't care about what this roon person on Twitter thinks and you'll just be glad you have a comfy chair for the joints
@giansegato
This is quite a simple answer.
Facebook built an LLM. They moved the team to France. That team quit and built Mistral.
So how did the French innovate? They didn't. The Americans did and jammed the innovative team down their throats.
Prediction: pure front end engineering will be fully replaced by AI within 5 years.
I know a few startups, such as
@coframe_ai
, that have no front enders and already 100% generate all FE code. The lowest skill area of engineering will be the first to fall.
Today we're excited to introduce Devin, the first AI software engineer.
Devin is the new state-of-the-art on the SWE-Bench coding benchmark, has successfully passed practical engineering interviews from leading AI companies, and has even completed real jobs on Upwork.
Devin is
@giansegato
Lol. The answer is a lot less exciting than you think: Company politics, organizational battles, and office space. FAIR already had one division in France that needed more ownership.
Source: I was there
@deedydas
It's always OK to replace Asians.
They have the fortitude to outwork everyone to overcome setbacks from discrimination, which means they're privileged or something which in turn makes it ok to further discriminate.
At least that's the commonly accepted rhetoric.
Inflection CEO quitting is a sign.
VCs are going to lose a ton of money chasing these high valuations from founders looking for quick money. Getting pitched a 1.5b preseed that someone actually did is as much of an indicator as $dogwifhat
Noob mistake: lowering your valuation to attract investors.
Valuation is rarely, if ever, the problem.
The problem is: are you an attractive deal? Are you the best deal Johnny VC has seen in the past 3 months?
Spoke to a friend who's a Partner at another Tier 1 firm.
His major complaint was that the firm as a whole doesn't understand AI and treats it as "another hot fad," so no one really spends time to go deep.
This is a tier 1 that EVERYONE has heard of.
🤯
As a former YC admit and current Partner at Khosla Ventures, I'm happy to review YC apps and help!
I don't have time for everyone, but DM me your application and I'll try to give some feedback.
I've coinvested with
@Lux_Capital
and we love working with them at Khosla.
The fact that
@wolfejosh
and
@bznotes
can hold opposing views and still enjoy working with each other is why they're both class acts and is an example of how all of us should behave.
@volodarik
If Pitchbook is to be believed, founders and employees generated 0 shareholder value so were compensated accordingly.
This doesn't sound that nefarious. Looks like investors invested close to 465m, so a near 1x liq pref is pretty damn clean terms.
@Carnage4Life
The average bootcamper graduated as a 3rd rate engineer not worth their salary until they had 2-3 years of experience under their belt. Effectively companies paid them to give them an education.
Yes that is the definition of a ZIRP level investment
I learned today there's specific VC partners at well known firms that in crowd founders know give high valuations and have no value add beyond being likeable people. It's well known to throw out any advice they give.
Guess this works in zirp. I wonder if it still works today.
@garrytan
Having been a part of 2 decacorns and 1 unicorn from early to late stage, I'll say that the founders are usually not the manipulative ones. It's the hired in VPs that are the problem.
@ylecun
@vkhosla
I'm not sure I follow
@ylecun
. A human's lifespan is finite and we can only talk so quickly. So by definition, humans also have a finite number of tokens they can produce.
Are you then claiming humans cannot reason (although I have questioned this for some humans 😛)?
@anothercohen
I sympathize, but the fact remains she didn't close anything. And others in her onboarding cohort almost certainly did.
This is an AE position. The only thing that matters are deals closed and size. Luck, bad luck, skill, etc are all irrelevant next to that number.
You only got 60 post for a preseed and wanted 100 so you throw a fit and threaten to not start the company? You aren't founder material.
Folks, don't work for founders who are driven by extreme greed.
Don't care if they're a PhD new grad from a hot AI lab.
Jokingly agreed you should never pitch associates, so everyone started pitching me yesterday instead. Ended up responding, reading, and giving everyone feedback.
Final tally:
56 total pitches
14 deeper dives over DM
3 followup Zooms and potential investments!
Spoke to someone interested in startups that said they'd rather join Pika Labs at a 1B valuation than the Meta of today at a 50m valuation.
Folks, the name of the game is to buy low and sell high. You want the valuation to be low when you join and spike AFTER you join.
FANG employees, I encourage you all to start companies.
But if you tell me you're raising 8m for a seed and your operating plan assumes all employees cost 500k-1.2m, I'm going to pass and let you learn how to be scrappy on someone else's dime.
@BarackObama
@POTUS
@BarackObama
I'm truly disappointed. I'd encourage you to consume material that's actually about AI and understand how the technology functions and what it's true implications are. At a glance, none of the authors of those books are credible enough to speak on the topic.
@eshear
A refrigerator with no bells and whistles and the coils vertically placed on the back so it'll last 40 years vs 5 years.
Nobody makes them anymore because it's not profitable. You need things to break to make money 🙁
The YC batch graduated should celebrate and have fun for a job well done.
But I can't help but feel nostalgia for the time where the yc class would pour one out after DD then turn back to their laptops and code instead of partying and clubbing. Sometimes it feels like finance
@ekzhang1
Not that niche. I graduated in 2011 and every CS student from a top school was aware of all of the competitions (ICPC, IMO, IOI, Putnam, etc). It's been a signal for top tech firms as long as I've been around
Sometimes, a founder and product just blow you away. Excited to be an investor in
@cognition_labs
and lucky to have been part of the team to make this happen.
Looking forward to working closer with you
@ScottWu46
!
Excited for Devin, the first AI software engineer, from
@cognition_labs
and happy to be participating in the financing. This was the CEO (
@ScottWu46
) 14 years ago:
Company I'm close to is upending the anime industry and already has contracts to remaster a couple Naruto movies with GenAI.
They're in dire need of a computer vision expert / CTO. Willing to give founder level equity for the right hire. Anyone interested?
Deepmind has a retention problem and someone is leaving to start a company every week. VCs, you don't need to pump up the valuation anymore. You'll have another shot on goal next week
Founder networks are one of the greatest support networks when starting up (thanks YC /
@paulg
!) and this is why founders from prior eras like
@vkhosla
or
@patrickc
can feel like they're "built different."
A whole different level of loneliness and mental fortitude was required.
@tmrohan
Agree. This was how it was at Palantir. There's something special about that environment and the learning that occurs through osmosis when you're young.
Thus is under appreciated. A lot of people ask me what to look for when joining a startup. I always say
1) is it going to break out?
2) can you trust the founders to not screw you?
Only veterans aren't surprised at 2.
Anyone shocked by OpenAI's equity clawback needs to understand that this sort of thing is way more common that you think.
When Microsoft bought Skype, Skype zeroed out the stock options AND stock of employees who left the company (and even from some employees who didn't)
When
Hot take: every CS program should only teach intro courses in C. People learn the fundamentals much better and end up being much sharper technologists long term
More first time founders these days tell me I "don't get it" when I ask about defensibility. "Google can kill any startup. That's a bad reason to not invest in us."
:s/Google/OpenAI/
OpenAI devday is a reminder that moats matter. Always have. Always will.
@Austen
It baffles me how often people reject candidates because of feedback like this. Hiring SWEs from coding interviews was one of the most meritocratic hiring methods before leetcode gamed the system
People don't give
@OpenAI
and enough credit for saving SF. Remove them, and SF would have no AI founder community upside to being here and in a continued decline.
Most young people who flocked to NYC would still be there.
@gokulr
Doing venture when you're young is a risk most shouldn't take. You learn skills that aren't that applicable to operating and there's a pretty sharp culling of the herd at each level. If you leave VC you end up with a skills gap compared to your cohort
Google employees are actively occupying the California office of Google Cloud CEO Thomas Kurian and refusing to leave until the company stops doing business with Israel, video live-streamed by the employees shows.
My latest:
@clairevo
They also bring in massive politics and almost always think individual first vs company first. Ymmv but every statement for and against consultants here seems to paint an accurate stereotype.
Hire for what you value and need
I think I just witnessed history.
AEO (answer engine optimization) by
@natfriedman
. The dude literally tells an AI agent to call him handsome on his website. And
@perplexity_ai
does! This is awesome (and kinda terrifying)
@AravSrinivas
.
@itsurboyevan
@pitdesi
@kwharrison13
@FoundersPodcast
Your moral fiber requirement is a poor filter. You're first filtering to 1) your morals vs other than arguably better morals 2) people you personally know or can tell stories about.
I know a few in that list that check all your boxes. You're likely writing them off because of
PSA: Great company != Great engineers.
People often conflate this and end up hiring B or C players from pedigreed companies that did well in spite rather than because of their eng talent.
I've had 10 pitches land in my inbox so far and one of them looks promising and could turn into an investment!
Happy to read every deck and provide at least 1 line of feedback
Everyone underestimates
@gdb
Algorithms are cool. But you can create ChatGPT with a ConvNet. Why Transformers? 10x improvement in training cost.
Guess what else is a 10x improvement?
Optimizing the shit out of CUDA.
You don't get ChatGPT without Greg and the systems folks.
The current HF0 batch is our best yet:
- a decacorn founder starting their next company
- a former openai researcher who sold her last AI startup
- and three teams that already broke $1M annualized revenue
If you know someone we should consider for our next batch, intro us.
@deedydas
PM had become non technical and deteriorating as a role since at least 2014. Companies could just afford to not fix the issue during ZIRP and it's not like the product org didn't deliver value. It's just they the value came from 1% of the people so it takes a lot of effort to
@then_there_was
You're punishing the man for avoiding the limelight. All of his quotes are from his early 20s
"I'm CEO, bitch"
"young people are just smarter"
"move fast and break things"
Now, he just avoids controversy. Work with the guy and you'll realize FB's dominance is because of Zuck
@letannock
@Carnage4Life
I don't think you understand how this works. They don't work for YouTube.
Google hired a firm to do one time rote labor. That firm employed people to do it. This union is like McDonald's threatening to raise prices because I'm buying a burger and not fries. Fine, I'll just go
@deedydas
@pitdesi
The major difference is that OpenAI actually had something to bet on: a research lead, a massive talent lead, an unproven market with no incumbent.
Chasing a leader in a competitive market is very different than a blue ocean
Interesting observation from older execs and VCs on the west coast.
Their kids grow up in a bubble and start liberal. But in college, a lot have a sharp change towards conservative.
The trigger?
Internship -> paycheck -> Tax bill
Closed distal fibula fracture aka broken ankle. My appreciation for the ability to walk is increasing now that I can't do it.
Surgery Monday with metal plates and screws 🥲
The team is 🔥. This team has no frontend eng, generates all their React code with an LLM, and is now releasing some of those tools for you to use. Check this out 👇
@realBobbyHealy
@paulg
@vkhosla
Yes exactly! I've had this same conversation multiple times. The amount of arrogance required to think you can stop progress is baffling.
@shreyas
Nah. It's because most companies have no idea wtf a pm is supposed to be and hiring for PMs is fuzzy. Someone inevitably gets a fancy title and uses that to hop elsewhere as a VPP all without ever having done real product work.
Here's a sad hot take. 50% of senior engineers and
@BillAckman
@MIT
Sad to see MIT here as they aren't a liberal arts university.
Maybe it's time for
@CarnegieMellon
to step up and be the USA's forefront technical university.
fresh new collab brewing
we at
@anduriltech
have been working with
@PalantirTech
to deploy Foundry internally with an initial focus on streamlining our S&OP process
look at that acceleration 🤤
Hot take: flat taxes like Singapore would be more progressive and a net positive for the US.
I bet the top 0.1% ends up paying more with flat tax and we can do away with all the corrupt "tax breaks."
And how much will GDP increase when we work instead of waste time on taxes?
@blader
Meh. I used to think this way and then became a VC. Everyone should just be an adult and stop playing psychological jiu jitsu games. You're probably not very good at it and the VC is more practiced.
If a good VC can't make time for you on the weekend, they may have kids, have a
I've seen so many leaders fail to understand this all the way to VPs at FANG.
You may not be a domain expert, but you can test for high quality inputs. This is why
@VSwordH
is world class and we love working with him at KV
@notgregkalai
I mostly try to understand the reasoning behind something. For example, did they think about second-order effects, what did they optimize for, and what compromises did they make to get there? In a nutshell, is this a half-assed job or the byproduct of a good thought process?
So your co wraps am LLM and investors keep asking about defensibility! What do you do?
What companies have always done. Build a traditional moat. Viral network effects, workflow integration, data gravity, etc. Don't forget about the well trodden path because of the new shiny
@patio11
@nearcyan
Ooo! I can do this one.
"tokyo is able to squeeze in more units because they don't have SFH, yards, or regulations like fire safety and handicap accessibility allowing smaller apartments"
@bcantrill
Teacher recs are also highly biased and can vary for the same student if they go to an elite institution that knows how to judge them vs a school in the boonies that's racist towards their demographic.
Protip: if you're joining a startup, ask how much of the employee equity pool is available on a fully diluted basis.
The larger it is, the more management prioritizes rewarding employees and the less likely they'll need to dilute you out in the future.
@jasonlk
I've seen worse. Had a 1.5B pre seed cross our desks at Khosla a few weeks back. The guys hadn't even left their old FANG jobs yet.
Someone actually did it.