Jhonatan Sanchez

Founder Agencia de Automatización IA

[Yarchi @undefinedKi]

THIS GUY CONNECTED HIS AI AGENTS TO HIS OBSIDIAN AND BUILT A BRAIN THAT LEARNS ON ITS OWN. HERE'S HOW TO BUILD IT

Obsidian is just markdown files sitting in a folder. That turns out to be the perfect memory for an AI agent, because an agent can read and write those files directly. He wired his agents into the vault so they pull context from it, do the work, and write what they learned back. The notes aren't the point. The loop is, and it gets sharper every cycle

How to build it:

1. Point an agent at your vault. The fastest way, no plugins, no API keys: open a terminal and run npx obsidian-mcp /path/to/your/vault. That exposes your Obsidian folder to Claude as a tool it can read, search, and write to. Add it to your Claude Code or Cowork config and restart

2. Confirm it can see the brain. Ask it: "list the notes in my vault and summarize what's in them." If it reads them back, the connection is live. Now it starts every task with everything the vault already holds instead of from zero

3. Give each agent one job and a write-back rule. Tell it: "research this, then save what you found as a new note in /brain with links to related notes." One agent researches, one summarizes, one plans. Each writes its output back into the vault

4. Close the loop. Add one line to every agent's instructions: "read /brain before starting, write your result back when done." Now each task leaves the vault richer, and the next run reads that before it works. It compounds instead of resetting

5. You only steer. Review what the brain produces, point it at the next thing. The agents handle the reading, writing, and connecting

The edge isn't better notes. It's a brain that feeds itself, so the work gets sharper every cycle instead of starting over

Bookmark this

[Utkarsh Sharma @techxutkarsh]

🚨 BREAKING: Claude can now grow your Instagram page like a $15K social media manager for free.

Use these prompts to get more views, followers, and sales in 2026.

(Save this before it disappears):

[monokern @monokern]

Claude Code + NotebookLM + Obsidian. A research pipeline that runs itself and remembers everything

The problem isn't finding information. It's that none of it sticks.

3 layers that fix this:

> Claude Code runs the pipeline on one command - no manual steps

> NotebookLM processes up to 50 sources on Google's infrastructure - not your tokens

> Obsidian stores every result and turns into a knowledge base Claude Code can actually read

After a year of consistent use you have a system that has seen hundreds of your research sessions, knows your conventions, and delivers outputs in exactly the format you want

Without prompting for it

Setup: 30 minutes
First session: 6 minutes
Payoff: compounds indefinitely

full guide in the article

[monokern @monokern]

https://t.co/RmlXgqYeJg

[Tran Mau Tri Tam ✪ @tranmautritam]

Paste a URL. Get a DESIGN. md

[Viktor Oddy @viktoroddy]

Claude Design is insane.

❤️‍🔥Just recorded a 18-min tutorial on how to build animated, award-winning websites with Claude Design + Opus 4.7!

[Charly Wargnier @DataChaz]

🚨 SKIP THE NETFLIX BINGE TONIGHT

Give this an hour instead, and it will forever change how you approach Agents and Vibe-coding.

@bcherny the Head of Claude Code doesn't hold back.

tbh he's delivering more value in 90 minutes than most AI paid courses 💥

[Riley Brown @rileybrown]

yup.. Someone made a Figma for AI Agents

Claude Code and OpenClaw can now create designs 10x easier...

00:00 Intro
00:46 Installing @paper
01:56 Connecting Claude Code to Paper
03:38 Plan for Design
06:23 Claude Code designs
08:28 Iterating
11:48 Building the React app
13:37 Deploying

[Julian Goldie SEO @JulianGoldieSEO]

Google Antigravity FULL COURSE 2 HOURS (Build & Automate Anything)

[Charly Wargnier @DataChaz]

Vibe-coders, bookmark this.

@YCombinator breaks down how to level up your vibe coding 👇

[Julian Goldie SEO @JulianGoldieSEO]

Instead of watching a 2 hour movie on Netflix, watch this masterclass on Claude AI and thank me later.

[Daniel Ch @chddaniel]

IT'S SO OVER!!! Claude Opus 4.6 can one-shot anything.

Prompt below 👇

[Daniel Ch @chddaniel]

I tried this prompt on @shipper_now (runs on Claude Opus 4.6)

" Create a personal portfolio landing page for [YOUR NAME] built as a single full-screen hero that occupies the entire viewport. Use a large portrait image [IMAGE ONE] as the background, centered and scaled to cover the screen. Place my name in the top-left corner in an elegant serif font such as Playfair Display, with the first and last name stacked on separate lines. In the top-right corner, add a simple navigation link labeled 'F1 Records'.

In the bottom-right corner, include links to Instagram and X/Twitter using solid, filled SVG icons so they remain clearly visible and easy to recognize. Each icon should link to the appropriate social profile.

The primary interaction should be a circular highlight cursor that follows the mouse smoothly with a slight delay. This round spotlight acts as a reveal mask, exposing a second version of the portrait [IMAGE TWO] beneath the main image. As the user moves the cursor, it should feel like they are scanning across an alternate layer of the photo.

When the cursor moves faster, the circular highlight should leave behind soft, fading echoes that quickly dissolve, adding motion and responsiveness. In the background, add a subtle animated grid pattern that gently reacts to cursor movement, creating depth while staying understated.

All text elements, including the name, the “F1 Records” link, and the social icons, should dynamically invert to white when the circular highlight passes over them to maintain contrast. Use smooth transitions of around 300ms. Apply a light parallax effect where elements shift slightly opposite to cursor movement. The overall look should remain minimal and refined, with a clean white base and the imagery and interaction as the visual centerpiece."

[Jhonatan Sanchez @Intellexia_io]

@Motion_Viz System

[Meng To @MengTo]

How I created these landing pages with Gemini 3 from start to finish

First, I start with the hero section. It includes the nav bar, eyebrow, headline, subheadline, cta, social proof and visual. I spend 50% of the time here because it sets the colors, typography, spacing, which AI uses for the rest of the site consistently.

“Create the hero section for my {app} called {name} in the style of {reference_site}”.

Pro tip: use a screenshot and you’ll get way better results.

Let’s get into the details.

For icons, I prompt: “Use Iconify {icon set name}”. Most people use Lucide, but there are hundreds of open-source sets on Iconify like Solar, HeroIcons, Iconoir, Phosphor, etc. Just need to mention in the prompt. Same for custom fonts.

For the animation, I prompt: “Animate fade in, slide in, blur in, element by element. Use 'both' instead of 'forwards'. Don't use opacity 0.”. This creates a subtle intro animation the first time users land on your page.

Gemini 3 is an excellent animator. For example, I created the beam animation with this prompt: “Add noodles that connect and beam animate into the right circle. Add subtle details to the right beam animation circle with sonar and decorations.”

For background animation, I use Unicorn Studio. Remix one of their templates and watch people click on your cover like crazy. What can I say, people love lasers.

In the hero or right below it, social proof is super important. You can put ratings or logos, or both. For logos, you can prompt: “Animate the logos with marquee animation looping infinitely using duplicated items and alpha mask.”.

Now the CTA. AI always creates basic buttons, which is fine 99% of the time. But Gemini 3 now sets a high bar for baseline design, so you will need to stand out. That’s why I put the extra human touch on animations and lickable buttons. I suggest browsing UIVerse and Codepen for buttons and reference the code in your prompt: “Change main button {code} and secondary button {code}. Add a 1px border beam animation around the pill-shaped main button on hover.”.

Once you’re happy with the hero, you’ll want to craft new sections based on your business. Features, action plan, pricing, testimonials, FAQ, CTA and footer are the popular ones. Insert a screenshot and prompt “Adapt a new section, change texts, names and numbers”. Gemini 3 is very smart. It reads your existing styles and site concept and will tastefully mold new designs to fit perfectly into your current site.

Finally, repeat the same prompts for icons, buttons and animations. Use midjourney images and remix using Nano Banana Pro. Ask ChatGPT to come up with better headlines, features, ctas, etc. Don’t skip the human part, this is where you’re irreplaceable. Start prompting top-tier landing pages and watch your numbers grow.

[ℏεsam @Hesamation]

> be Gabriel Petersson
> Swedish kid
> buy and sell Pokémon cards
> make 20k before even finishing puberty
> still have homework tho lol

> realize school is side-quest
> drop out at 17
> join Silicon Valley as founding engineer
> learn machine learning by just doing it
> sell AI to companies by literally knocking on their doors
> startup grind is real

> become CTO at 19
> build engineering team

fastforward
> join Midjourney
> build world’s fastest image grid
> casually invent multithreaded canvas
> also tools for hyperparameter tuning
> also tools for dataset exploration
> also tools for things humans don’t understand

> “yo America let me in”
> visa man: “degree?”
>you: “no but look at my Stack”
> O-1 approved (extraordinary individuals)

2024
> join OpenAI Sora team
> research AI video models
> “make the AGI touch grass but virtually”
> talk about sims where AI rediscovers physics
> actual cutting-edge science
> all self-taught

[The Startup Ideas Podcast (SIP) 🧃 @startupideaspod]

How to set up Claude Skills in 3 min (for complete beginners)

Step 1: Create a markdown file with your EXACT process
Step 2: Add reference docs (brand guidelines, formulas)
Step 3: Write deterministic scripts for calculations
Step 4: Deploy once → Same output forever

Result: No more prompt engineering. No more hallucinations.

Tools: Claude Skills (not basic prompting)
Cost: $20/month
Time: 30 min setup → Save Hours of Prompting

You're one Skill away from building a 10x AI Agent

Use this system, you won't regret it

[Daily Dose of Data Science @DailyDoseOfDS_]

The open-source RAG stack (2025):

[Jhonatan Sanchez @Intellexia_io]

@aminaak68329056 @figma Me