shullllllz

#btc

[levithefirst @levithefirst]

if you are tired of your websites looking vibecoded here are 15 UI resources you need.

1 - https://t.co/T08XPa6D08 - design playbooks for humans and coding agents. can be used to audit UI, fix generic AI design, improve motion, accessibility, and polish.

2 - https://t.co/Mfyq9u2mhB - a https://t.co/88NNTSFSeL’s production-ready design system. Accessible React components you own. Great for building real product UI without the usual SaaS look.

3 - https://t.co/wJhLWQvcdy - it's a practical design-system checklist that stops agents from creating random UI styles.

4 - https://t.co/6D7ztIfoNZ -
1,000+ free shadcn components and advanced patterns like data grids, kanban, gantt, filters, and calendars. very good for complex product interfaces.

5 - https://t.co/e1sRqfnNqL - ready-made spring animations with CSS, React, and AI prompts.

6 - https://t.co/uvWP6nV2dI - free browser-based custom icon designer.

7 - https://t.co/iDFqVg7kc4 - 256 prompts for specific UI sections like dashboards, pricing, auth, onboarding, and hero sections. Give agents a concrete layout target instead of “make it look good.”

8 - https://t.co/oVf16brNhN - 99 CSS-only button interactions. Copy the CSS or prompt. Good for keeping interactions consistent across your product.

9 - https://t.co/guZnefI81c - 2,600+ examples from 95 real design systems.

10 - https://t.co/Lh6pTUglpP -88 production design systems with tokens, stacks, and downloadable design(.)md files.

11 - https://t.co/pEx4qJHw4Q - generates fluid typography and spacing.

12 - https://t.co/ZOclayT7xU - ready-made design tokens for colors, shadows, radii, spacing, easing, and more. Stops agents from inventing random values everywhere.

13 - https://t.co/95Bpzx3s8g - checklist of tiny interaction details that make interfaces feel finished.

14 - https://t.co/bs0bp88S90 - copy-paste backgrounds for Tailwind and CSS.

15 - https://t.co/dRMIceievX - reusable motion components for React.

[Agustín Antonetti @agusantonetti]

Por primera vez en la historia se juntaron Google, Microsoft, Anthropic, OpenAI y otrás 100 compañías más del sector para publicar una carta pidiendo a los gobiernos que se preparen para recibir ciberataques con los nuevos modelos de inteligencia artificial.

Fuerte.

[joshpuckett @joshpuckett]

Here's an actual design engineering tip:

Safari sucks at updating multiple WebGL video textures in the same frame, especially during interactions or animations on initial page load.

To keep things buttery you'll wanna pause texture uploads during motion e.g:

isMoving ? holdLastFrame() : gl.texImage2D(video)

[Nakul Mandan @nakul]

"10,000 hours? 996? That's rookie numbers. Like... what are we talking about?"

Guillermo Rauch (@rauchg). Founder and CEO of @vercel.

Only a founder as intense and energetic as Guillermo can call 996 "rookie numbers", and you still nod along. The man who's coming after every pixel and every token of the internet.

Knuckle Up ↓

00:00 Who is Guillermo Rauch?
02:01 What is Recursive Founder Mode?
07:28 Why did Vercel open source its playbook with skills dot sh?
10:14 What is a "work prompt" interview?
14:33 Why is Guillermo still involved in every hire at nearly 800 people?
18:17 What keeps founders at Vercel when seed money is everywhere?
23:41 Do urgency and taste pull against each other?
26:43 What is the new engineering meta skill?
32:21 How does a 17-year-old show they're exceptional today?
35:47 Why does Vercel route the whole company through one agent?
38:47 Which software survives the agent era?
42:11 How has AI changed Guillermo's job as CEO?
51:39 Does Vercel even need executives anymore?
54:40 If Guillermo started Vercel today, what would he build first?
57:45 What's the story behind v0?
1:03:48 Are there still moats when anyone can build fast?
1:11:24 How does a legendary coder learn the inner game of being a CEO?
1:19:21 Where does Guillermo's energy actually come from?
1:23:07 Quickfire: overrated advice, Chrome, voice AI, Demo Fridays
1:28:38 What does the journey to greatness actually require?

[shullllllz @irigoyai]

@jesper_alpacka @nuxt_js @greensock @threejs @datocms You rock, gonna check those out. Thank you kindly and keep blowing our minds

[Y Combinator @ycombinator]

AI-Native Service Companies
@gustaf

The total spend on services is many times larger than the spend on software, and a lot of those services are already outsourced, which makes them easier to replace with an AI-native product.

We're excited about companies that don't sell a tool to help you do the work: they just do the work.

[Google Ads @GoogleAds]

Wait… 1 prompt just made 10 on-brand ads? 🎬

Meet Gemini Omni in Google Ads. It’s your new go-to for generating, editing, and scaling on-brand videos directly in Asset Studio.

Ready to turn prompts into high-performing ads without losing brand integrity? Try Gemini Omni in Asset Studio rolling out now.👇

[Zack (Paid Ads Specialist) @zackpaid]

do you have ANY idea what this means for meta ads?

higgsfield just launched an ad multiplier that generates complete creative diversity for Meta Andromeda and it's crazy good.

you can now rebuild all of your meta ads winners without touching canva or hiring designers or having to think of new ideas you can build.

20 ad creative variations testing different visual styles, product photography mockups for service offerings, carousel ads with consistent branding across all slides, video storyboards and brand guidelines

the bottleneck has always been creative production speed.

you'd spend $200 and wait 3 days for a designer to deliver 3 variations.

now you test 50 concepts of proven winner ads in an afternoon.

here's some prompts for it in the replies under this post:

[ElBuni @therealbuni]

En china hicieron que unos robots jueguen futbol y uno metio terrible falta
Se lo tuvieron que llevar al otro
Encima despues seguía tratando de romper a los demas
Ese vino asi de programación, no e casualidad jsjs

[Adham Dannaway @AdhamDannaway]

Morph an SVG icon into any other 🤯

→ https://t.co/uftGVeJYlQ

By Guillermo López

[Erick @ErickSky]

Le tumbaron el github a este dev porque nos dio ABSOLUTAMENTE GRATIS una app para entrenar que no te pide cuenta, suscripción ni guardar tus datos en la nube y que es MUCHÍSIMO MEJOR QUE CUALQUIER OTRA EN LA APP STORE

[openGym]

Un tracker de entrenamiento self-hosted en el que puedes:

• Planificar tu semana
• Seguir entrenamientos guiados
• Registrar cada serie, peso y repeticiones
• Ver PRs, estadísticas y evolución
• Registrar peso corporal
• Crear superseries
• Seguir progresiones automáticas
• Analizar qué músculos estás entrenando
• Usar passkeys / Face ID / Touch ID
• Funcionar offline
• Sin anuncios
• Sin tracking
• Sin suscripción
• Sin depender de los servidores de una empresa

Además, tiene 1.324 ejercicios, importa datos desde FitNotes, Strong y Hevy, y permite exportar todo en JSON.

REPOOO👇

[Guido rnr @GuidexOrtigoza]

El último video que voy a ver antes de morirme

[CopyRebeldia @CopyRebeldia]

Así se ve un sueldo de 750.000 dólares al año: un tipo en camiseta blanca, un pizarrón y 2 horas y media.

Stanford, CS336. Percy Liang construye un LLM desde cero. Lo que hay debajo de Claude y ChatGPT, y arranca por la parte que todos saltan: el modelo no lee tu texto, lee números.

Anthropic paga ese sueldo a los ingenieros que entienden esa capa.

Lo único que cobra Stanford son 2 horas y media de tu atención.

[Dividendology @dividendology]

This image has gone mega-viral over the last 24 hours.

If you bought Micron every day at market close and sold at market open, your return would be 138,330,342%.

If you bought Micron every day at market open and sold at market close, your return would be -99.2%.

I went down a rabbit hole to see if I could discover as to why this is.

An earlier historical dataset from researcher Bruce Knuteson produced similarly extreme results for a number of individual stocks:

• $MU: +16.2 million% overnight / -99.95% intraday
• $HOLX: +15.9 million% / -99.98%
• $NVR: +3.6 million% / -98%
• $JBHT: +3.0 million% / -99.8%
• $MGM: +2.6 million% / -99.96%
• $NVDA: +221,715% / -51%

Obviously, the exact numbers depend heavily on the period examined, but the pattern itself has been repeatedly documented.

Research published in the Journal of Investment Management found that the phenomenon is particularly strong among high-attention stocks popular with retail investors, including names such as:

$NVDA
$TSLA
$AMD
$MSTR

But here's what I didn't expect:

Over the 30-year period examined in subsequent research, $1 invested in only during regular trading hours grew to roughly $1.20.

The same $1 invested only overnight grew to approximately $17.27.

So why does this happen?

There are several theories.

One of the most interesting involves when investors actually place orders.

Retail investors often place orders during post or pre market hours.

Those orders accumulate while the market is closed and are executed around the opening auction.

That creates persistent buying pressure near the open, especially in popular, high-attention stocks.

There are a few other possible explanations as well:

• Earnings and major corporate news frequently arrive outside regular trading hours.

• Investors require compensation for accepting overnight gap risk when they cannot immediately exit a position.

• Institutional investors face different leverage, short-selling and financing constraints overnight.

• Liquidity and price impact behave differently around the opening and closing auctions.

• High-beta stocks appear to experience a particularly large day/night return split.

Researchers still disagrees on how much each explanation contributes.

[Massimo @Rainmaker1973]

Exercises for neck and shoulder relief

[Gonzalo Bauzá @gbauzap]

Es simple:

Si tu SaaS o App tiene búsquedas en orgánico → hacé SEO.

Si tu SaaS o App resuelve un problema visual o aspiracional → hacé contenido.

Si tu SaaS o App tiene un dolor urgente y claro → probá anuncios.

Si tu SaaS o App mejora con más usuarios → pensá en referidos o viralidad.

Si tu SaaS o App vende a empresas → hacé outbound y linkedin.

No todos los productos crecen por el mismo canal.

El error es copiar la estrategia de otro sin entender tu propia palanca.

[Will @willbakerships]

hello again

[Frederick James @frederickjames]

Yesterday, I resigned from my job.

I also hit $10,000/month.

It took me 88 days.

Keep going friends. The world is yours.

[Nav Toor @heynavtoor]

Take any song on Earth. Any artist. Any era.

Upload it. Click one button.

The AI splits it into individual tracks. Vocals in one file. Drums in another. Bass in another. Guitar in another.

The Beatles. Kendrick. Taylor Swift. Billie Eilish. A cassette your father recorded in 1974.

Every instrument separated. Clean. Individual. Playable.

Studios could not do this cleanly ten years ago. If you did not have the original session files from the mixing engineer, the vocals and instruments were fused together forever.

Then in 2020, two pseudonymous developers who go by Anjok07 and aufr33 uploaded a Windows app to GitHub. Free. MIT licensed. No ads. No account required.

It is called Ultimate Vocal Remover. UVR5.

25,833 stars on GitHub. And here is the number nobody talks about. One release, v5.6 from September 2023, has been downloaded 5,161,858 times directly from GitHub. The full release history sits at over 6 million downloads.

Six million people quietly downloaded one open source tool to do what the music industry sells as a premium AI feature.

Here is what it separates.

Two-stem mode. Vocals and instrumental. A karaoke track from any song in minutes.
Four-stem mode. Vocals, drums, bass, and other. Every rhythm section on its own layer.
Six-stem mode. Vocals, drums, bass, guitar, piano, and other.

Three AI engines built in.

MDX-Net for clean vocal isolation. Meta's Demucs v4 for multi-stem separation. VR Architecture for post-processing to strip reverb and echo from extracted vocals. Chain them together. Extract vocals with MDX-Net. Clean them with VR. Studio-grade acapella from a YouTube rip.

Here is who uses this.

DJs pulling acapellas for live sets without begging labels for stems.
Producers sampling drum breaks from 1970s funk records.
Music teachers isolating a bass line for students to practice against.
Karaoke makers turning any song into an instrumental track.
Podcasters removing background music from an interview.
Filmmakers extracting dialogue from a scene with a soundtrack.
Guitarists learning a solo by hearing it in complete isolation.

Here is what the paid alternatives charge.

https://t.co/otdeBPO0xL: $9.99 to $19.99 a month, with one-time top-up packs from $50 to $300.
Moises Pro: $29.99 a month on the US App Store.
iZotope RX 12, the post-production industry standard, sells its Advanced edition in the four-figure range.

UVR5: $0. Unlimited songs. Unlimited stems. On your machine.

Runs on Windows, macOS, and Linux. Works on CPU or GPU. NVIDIA CUDA and Apple Silicon both supported.

One honest note. The last GitHub push was March 2025 and v5.6 is still the current release. In most repos that would be a red flag. Here it means the tool works so well that 5 million people downloaded one release and never needed a v5.7. The upstream models it wraps around, Demucs from Meta and MDX-Net, are the state of the art in audio source separation.

Six million downloads. 25,833 stars. MIT licensed.

Every song is made of layers. UVR5 peels them apart.

(Link in the comments)