cabuuuu

[cabuuuu @caabuuu]

Vale la pena leerlo

[DAIR.AI @dair_ai]

If you maintain an AGENTS.md or a CLAUDE.md, this one is worth your time.

(bookmark it)

Researchers traced 94K development events across 557 agentic coding sessions, plus 690K file-level change records from 33K agentic pull requests.

Instruction files and working notes account for 60.5% of everything agents read.

Classical technical docs get 10.6%.

API references get 1.3%.

Reading docs is associated with less immediate testing, at an adjusted odds ratio of 0.39.

And consultation is self-initiated 70.2% of the time, against 7.5% driven by a failure.

In multi-commit agentic pull requests, code gets touched first 4.7x more often.

Paper: https://t.co/alrihmMkQd

Track more trending AI papers in our academy: https://t.co/LRnpZN7L4c

[cabuuuu @caabuuu]

@fedeeraizman @maxifirtman Filtraron que OpenAI iba regalar un año de plus a universitarios de EEUU. Si sale explica el Pro ahi. Afuera no se que tanto achicaron, ahora ofrecen menos pero a mas paises

[Pliny the Liberator 🐉󠅫󠄼󠄿󠅆󠄵󠄐󠅀󠄼󠄹󠄾󠅉󠅭 @elder_plinius]

Ox-alpha is from Zai, GLM-5.X family

my agent has spoken 🙌

[cabuuuu @caabuuu]

Ox Alpha es GLM. Le midieron el tokenizer contra el de todos los labs grandes y coincide con GLM en las 11 pruebas, ningun otro pasa de 4. Igual les conviene: con el nombre de https://t.co/S3JSvfGHTF puesto lo posteaban cuatro cuentas, como modelo misterioso te lo prueba medio Twitter gratis.

[OpenCode @opencode]

Ox Alpha (stealth model) is free for the next week

- 1M Context
- Multi-modal
- Zero Data Retention

Generous rate limits, near unlimited usage

We have capacity for 100T tokens per day, lets see what you can do

[cabuuuu @caabuuu]

@emollick Both can be true. They believed it and it was also the best fundraising line anyone had. Telling investors your product might be too powerful to control works great while nobody can check.

[Nick Davidov @Nick_Davidov]

I'll just drop this here

[cabuuuu @caabuuu]

https://t.co/GTcLxZzqk4
ejemplo de la semana: la metrica que eligieron para el lanzamiento es cuanto engaña al detector. de si el texto vale la pena leerlo no dicen nada.

[Ethan Mollick @emollick]

The problem is that the output from the AI detector fooling startup is bad writing.

[cabuuuu @caabuuu]

El estilo se volvio gratis hace un tiempo, cualquiera saca un texto prolijo. lo unico que sigue costando es tener algo propio para decir.

[Rob Hoffman @RobHoffman_]

I’ve been telling founders: you need to treat your content operation like a media org

if you do the same thing everyone else does (I.e rip Claude slop blog posts) you will get the same result everyone else gets

get your Inbound Engineer to extract unique data and information from customer interviews/surveys, sales calls, SMEs brains at your org

have them create a basic “study”

use it to not only rank but put out valuable, unique content on socials that will actually get you customers

the hard thing is the easy thing in the long run because the hard thing will make you rich

or you can keep ripping Claude slop like every other founder and watch your net-worth drop faster than your website traffic

[cabuuuu @caabuuu]

Asi tratan a los creators que les hacen el marketing. Al usuario comun peor, se quejan de los limites en el discord todos los dias pero renuevan igual. Mientras no haya alternativa mejor, tratar bien a la gente es opcional

[madpencil_ @madpencil_]

They representat Shit!
and their work ethics are shit.

I was CPP with Higgsfield paid posts, part of their CPP, all of it...

That was before few months. Deleted my Higgsfield account, and left their CPP Discord. I haven't been posting them for months anyway and avoiding communication.

So then, days before Kling 3.0 went public, while I was in the alpha test group, their Head of X Marketing came to me with a "confidential request": would I hand over my Kling 3.0 login?

$10k. Then $15k. Then $20k. Paid upfront. For "24 hours of internal access." They wouldn't leak it. They'd even "use a VPN of my country." Keep it confidential.

I said no. Politely. Every single time. My hands were shaking while replying to them was very nervous.

Walking away from $20k just for a login probably sounds insanely dumb to some people. But my access was given to me in trust, and that isn't for sale.

I reported it to Kling's officials the same night.

You're probably wondering why I'm sharing this now. I could have posted about it few months back when Higs was hot on pan just a couple of months back.... I avoid drama, I like to stay out of all these X drama like any normal creator will. But I saw a post today defending Higgsfield, so I figured my personal experience was worth adding.

They'll likely stay unbothered, and plenty of people will keep using the platform. And that's fair, the product itself isn't bad. It's the team running it that's been shady.

[cabuuuu @caabuuu]

@maxifirtman Lo que más me llama es eso que decís de que lo mantienen oculto. Si es legal, y encima destruir el ejemplar es parte de lo que lo hace legal (queda una sola copia, cambio de formato), ¿por que mantenerlo oculto? ¿es tema de imagen o hay algo mas?

[cabuuuu @caabuuu]

@zodchiii

[Stitch by Google @stitchbygoogle]

Today, we’re open-sourcing the draft specification for DESIGN.md, so it can be used across any tool or platform. We’re also adding new capabilities.

DESIGN.md lets you easily export and import your design rules from project to project. Instead of guessing intent, agents know exactly what a color is for and can even validate their choices against WCAG accessibility rules.

Watch David East break down this shared visual language in action👇. New capabilities and links in 🧵

[MatLab crashes @memecrashes]

any ai startup today

[Cristian Córdova 🐧 @barckcode]

🔥 Fuaaa van a habilitar un x2 en tokens en días laborables fuera del horario de 5 a 11 a. m. PT / de 12 a 6 p. m. GMT

Y x2 durante todo el día los fines de semana

Nada, a quemar tokens, disfruten!

[Brais Moure @MoureDev]

Aquí nadie se pone de acuerdo... Miles de CEOs acaban de admitir que la IA no ha cambiado nada en sus empresas

Fortune publica un estudio del NBER sobre 6.000 CEOs y CFOs en EE.UU., UK, Alemania y Australia. El resultado: el 90% dice que la IA no ha tenido impacto real en productividad ni en empleo en los últimos tres años.

Como siempre, los extremos no ayudan.

[cabuuuu @caabuuu]

ya parezco castre twitteando boludeces
perdon

[cabuuuu @caabuuu]

Lcdm 🚩🚩🚩

[cabuuuu @caabuuu]

entro a twitter y me encuentro a raizman hablando con gente de 60 años JAJAJAJA