Playing from timestamp
Video starts at 19:38
Channel
Interviewed Person
Malte Ubl
"Claude Sonnet 4 scores a 64. So like basically 100 minus 64 is the percentage that the apps that it generates just plain like objectively don't work."
"There clearly is a lot of value in being just a GPT wrapper if you do the user experience right."
"v0 is literally I mean it's very sophisticated now but like we just published a blog post about how its AI works but it's essentially an unchanged frontier model."
"The LLMs were better at doing the unified reasoning of HTML and CSS than they were at like writing a CSS file and then you know half a minute later doing HTML."
"What GMO does, our CEO, he takes screenshots of the Zoom call and just tweets that out. And if people tell him it's great, then it's on the roadmap."
[Music] Right. Welcome everyone. Really nice to see all of these familiar faces as well as the new ones here. I'm really excited about this chat. Ma, thank you so much for being here. Yeah, thanks for having me. And uh we have one question that we start off each fireside chat with. Um, it's a really general question uh that
we have for every of our guests. How did you get into computer science at all? Oh my goodness. Uh, do I have to say the year and age myself? No, no, no. Like I So I am from a town north of Hamburg. And I got a job when I was 14 years old at a software company. And this was still when they were actually distributing floppy discs. And my job was to label them and print out the manuals and then package it up for the mail. And at some point that was right
around the corner when the very earliest websites were being developed. And while like everyone there was like a math PhD and you know the at the time they were actually working in small talk if you heard of that. Um and no one knew how to make web pages. So I kind of just picked it up. Okay. So really early on and uh that interest stuck with you. Um just giving a brief overview. Uh Malta spent about
11 years at Google and then his latest role as the CTO of Vercel that most of you probably know. And um I want to talk first about your time at Google and kind of what got you there because um going uh to such a great company which I guess most of us kind of aspire and have these uh really big grand uh opinions about. Um what kind of got you there? uh how did you I mean fast forward from uh at 14 years old to your position there but
um kind of what led you there uh especially also going to the US uh instead of kind of staying here? Yeah. So like the the I actually so it's actually really funny story. So I started at Google in Hamburg, Germany and um as this was in Germany, you start I think always on the first which uh is like a random day of the week which Americans think is weird. Um but it was like so I think it was a Friday and I had my first ever of my life video conference. So this was like in 2010 and
uh and I was very excited and in this video conference I was told that they would close the engineering office in Hamburg, Germany. Um that was later at 4 p.m. my first day. Um and then I I I so but you know I wasn't like being let go. I was just being given the choice to to find a new team. And in hindsights this was really cool because when you first of all I mean I you know I didn't have any choice
at all what to work on in Hamburg. Obviously they were working exactly on one thing which was um keeping the my SQL database at the time that was running Google Adwords alive. And so now I you're on the inside, you understand like what the what everyone's working on, what's exciting and so I got to basically pick a team and uh and then I you know I I I ended up picking a team that was in Mountain View at the headquarters and essentially just stuck out the year in Germany to qualify for
L1 visa and then moved over. Uh which team was it that you ended up working on? So uh that tells you a lot about me being smart. Um it was uh the team that would eventually become plus. But the the lucky thing is that because it was so well funded um we I was working on on front end infrastructure. And so the we were basically we just had too much many people and and would would be able to build really cool things. And so the
