gianpaolo carraro

@fedemolina @rickyrauch @evilrabbit_ that’s because Claude Design does not drink Diet Coke

Claude Design is good, but my benchmark is @rickyrauch / @evilrabbit_ level taste. I still think we are years away from any AI model consistently reaching that bar, and I'm not sure it's even possible.

@mgonto @zachbruggeman @tryramp 99% men?

Today @zachbruggeman talked to us about @tryramp inspect, how they built it, how they use it, how they worked on the culture, etc. Another great Neurona Argenta meetup!

🤣

@yenkel AI was not there? looks like you are not a simulation believer 🤪

agree with the takes here I’d also add that your gut knows about the real value of things, like cooking for your family over precooked meals, or writing yourself vs with AI trust your gut. it’s developed for millions of years and AI wasn’t there

@yenkel via SAFE? they looking for more? you doing SPVs?

primer angel check de 2026 founder AI native, i.e. jugaba con AI antes de chatgpt, va contra una industria osificada en LATAM 💪 que lindo lo que se viene...

@fedemolina @v0 i just opened Los Tres Patos fun 🥳💪👍

Try out my Sunday Project: Coffee Tycoon made entirely with @v0 v0-coffe-tycoon.vercel.app

@yenkel @stripe betamax

not again 😆 is this (UCP vs ACP) like MCP and A2A? will @stripe support UCP?

@fedemolina @yenkel "All your base are belong to us”

I started using Claude Code today via the terminal in MacOS (thx @yenkel), and I am already at the point I’ve basically handed my entire OS over to Claude Code and just keep clicking “yes.”

@theHankTaylor @mgonto so in your view distribution > product ?

@gcarraro @mgonto any marketer worth their salt knows that product is a pre-req. But it's also known that an average product can win over a great product if the average product has better distribution strategies

without a good product doing sales or marketing makes no sense. cluely is a great example of how great distribution with no product makes absolutely no sense. having said that now if you have a good product idea, it's much easier to implement it. if you have a good product idea and you're able to implement it easily, distribution matters a lot more than before. @evilrabbit_ agree?

@yenkel @thdxr it does not have to… but that model worked pretty well for a while 🤪💪

@thdxr goes a long way to showing that the enterprise version of a product does have to be the consumer/pro + sso, SLA and support

remember that anthropic doesn't sell claude max to teams of course you can buy it for each of your employees but any serious company is going to want the teams functionality and they pay per token which is very profitable great loss leader setup

@mgonto then you think like a product person

@gcarraro I'm mostly marketing and I say product is more important

capturing Maduro feels a bit like:

@fedemolina “just” need military not wanting to die for maduro … which is plausible and testable by cia ahead of time

Como entras a un país y capturas a la persona más importante sin bajas significativas? Si algo me enseñaron los videojuegos que esto de seguro estaba arreglado no? Inside job o arreglado con Maduro de alguna manera

i’ve read many “my year in review” these past couple of days, thanks for sharing, here is mine: - enjoyed life

@mgonto @thomaspaulmann the rest quit between Feb 1 and Dec 30 , as you can quit only once, these % should be mutually exclusive and therefore can be added to 100%

@thomaspaulmann and the rest?

New Year’s Resolutions: - 23% give up by week one - 50% quit by end of January - 9% stick it out all year Pick wisely.

@mgonto @fmendiguren me too 🤪 🥳

@fmendiguren estoy. decime cuando y estoy para hacer solo viñedos en salta / Jujuy

Si alguna vez tienen la oportunidad, no dejen de venir a este paraíso imposible de transmitir en imágenes. (venir en 4x4 sí o sí)

@fedemolina @yenkel @martin_casado this shall be known as “the duck law” 🦆🤪

@yenkel @martin_casado this was always the case anyways with startups, only accelerated now by AI

I’ve found it interesting to log where spend time when coding with AI. The tl;dr is way less time dealing with library config and debugging platform (build, repo, deploy) issues. Way less time worrying about architecture up front. And way less time debugging simple logic bugs. OTOH, Way more fraction of time debugging bugs where the semantics are decoupled from the syntax (the more state I need to track, the less AI helps). Way more time refactoring. Way more time profiling and doing second and third order optimizations. And of course a lot more time thinking through and sketching out what I want to build out next. The 80/20 rule is definitely in effect. While going from 0 to a working project is 50x faster, it reduces to 2x faster or so when implementing and debugging complex stuff that requires trade offs based on runtime characteristics. This somewhat reflects the speed ups I see across the portfolios I work with. Those building distributed infra that requires strong guarantees and performance trade offs aren’t moving nearly as quickly as those closer to the app layer.

@yenkel tattoo next 💪

finally got to set up this reminder on my desk. remember ❤️

@jfroma 😎💪

@gcarraro By weather station I mean this kind of device

raspberry connect is so good! raspberrypi.com/software/conne…