Lanzamiento del proyecto "RD en la Prensa Norteamericana" versión 1.0
Las banderitas representan el idioma de su preferencia
Pueden leer el artículo - una copia verbatim de la guía dentro de la página - o meter mano de manera inmediata
(Buen provecho!) https://t.co/I6EvqFZhwW
@sethladd@Gemini_Notebook ah, cool - thanks
actually made a powerful video on Harnesses, Loops, and Graphs with one of my notebooks
you guys are extra busy up there so don't blame you if you don't watch it, but kudos on the API, this was an absolutely genius move
I’m at the office and this new dude is blasting Dembow at full volume
Manager isn’t saying anything to him, either
This is exactly why the Anglos want to deport you bro
It's wild how, if you take even a short break from this app, 3 days will suffice...
When you come back, the experience is something like being in an insane asylum with hundreds of people screaming their neuroses as loud as they can.
Tobi Lütke (@tobi), Shopify CEO, at Vercel Ship SF: "we are now all verification engineers. We are building harnesses, we are building acceptance tests, and we are building… verifiables."
When agents write the code, the work moves to the checks:
→ harnesses that make agent output repeatable
→ acceptance tests that define "done"
→ anything a machine can verify without a dev squinting at a diff
Full fireside with @rauchg in the reply 👇
First phase of Parque Colón and the Cathedral in the Colonial Zone of Santo Domingo
Nothing impressive yet but proportions are stellar
Using:
• Google Carbon 4 (High)
• Unreal Engine 5.9 + MCP
• Blender (Headless)
I mean, I dunno... could just be a random fire.
...but you have to admit it's only a matter of time before a frontier model cracks a pattern no one thought possible and something like this happens mysteriously
LLMs like to over-explain, I found this to be helpful https://t.co/O8vnM2xnRH
No one wants to navigate a website where the entire thing is a mile-long scroll of insanely elaborate explanations and multiple layers of redundancies, headers, and subheaders in places that are counterintuitive - but that's what it loves to spit out (because it is thinking in discrete mathematics and linear algebra)
Sometimes you have to beg the bot to make subpages and conduct multiple "adversarial and recursive reviews of the UX and UI" before it gets it right
"I understand that indexing information like this is useful for you as a calculator on steroids, but some of these things are already obvious to me as a human, please remember my sensory capabilities are very refined"
One additional thing that worked for me was taking a screenshot of one of its outputs, then using microsoft paint to primitively copy and paste things and move them around to brute force the design.
1. Then I had Claude do some image gen to polish it into what i was trying to express, and 2. then told it go off of the polished image as closely as possible - that's when it finally got it