Science Sets You Free

Unconflicted Science, that is. Occam’s razor applied to the cacophony of misinformation and propaganda. Collapse the dam brick by brick. Truth wins eventually.

[Anatoli Kopadze @AnatoliKopadze]

Anthropic engineers just showed how they build a full app from scratch, using a loop of agents

40 minutes from the team behind Claude Code

they used three agents: one to plan, one to build, one to judge, cycling until the app actually works

the winners won't have the smartest model, they'll have the best loop

watch it, then read the full guide on how to actually use loops below

[Ridark @ridark_eth]

I genuinely don't understand why everyone isn't using this yet

Andrej Karpathy, a co-founder of OpenAI, posted a simple idea that hit 16 million views: stop using AI to write code, use it to build a second brain.

You point Claude Code at a folder, drop in any source, an article, a transcript, a PDF, and Claude reads it, links it, and files it into a living wiki of everything you know. It compounds like interest, the more you feed it, the smarter it gets.

Here's the whole thing:

> Install Obsidian, create a vault, open it in Claude Code

> Paste Karpathy's wiki idea file and tell Claude to build it

> Claude makes three folders: raw for sources, wiki for its pages, a CLAUDE.md that runs it

> Drop any source into raw and say "ingest this"

> Ask questions across everything, forever

Five minutes to set up, and you never start from a blank chat again.

Full step-by-step guide with Claude and Obsidian, link below.

Bookmark this

[regent0x @regent0x_]

https://t.co/OKJUp3KAmo

[regent0x @regent0x_]

buys mini PCs from a factory for $600, sells them to law firms for $4,500 plus monthly support

the factory thinks he’s a regular bulk reseller

he’s running a $40k/month private AI business they have no idea exists

the factory rep assumes he flips them on amazon like everyone else who orders in volume

instead each box becomes a $50k client relationship over its lifetime
the arbitrage:

→ factory cost at volume: $600/unit
→ upgrades RAM to 128GB + storage: $200
→ installs local AI stack: 2 hours of his time
→ sells setup: $4,500
→ monthly support contract: $4,000/month

hardware markup alone is 6x

but the recurring contracts are the real business

his clients are who you’d expect - law firms with privileged files, medical practices under HIPAA, accounting firms holding client financials. all terrified of cloud AI, all carrying real liability if data leaks

he doesn’t sell them a computer

he sells the one answer that ends every objection: “this box sits in your office, your data never leaves the building, i can’t access it, the factory can’t, openai can’t”

the factory relationship is the wildest part

he orders 20+ units a month now. the rep keeps asking what retail channel he uses. he says “corporate accounts” and moves on

they’re assembling and testing these boxes thinking they’re shipping generic office hardware

they have no idea each $600 unit gets local AI installed and resold as regulatory infrastructure with a $4,000/month contract attached

the numbers after 8 months:

→ 10 active support clients
→ average contract: $4,000/month
→ recurring revenue: $40,000/month
→ cost per deployment: $800
→ his time per new client: one day

the support work is light - model updates, adding documents to the index, occasional troubleshooting. 6 hours a month per client

he keeps boxes stockpiled in his garage. referral comes in, he grabs one off the shelf, configures it overnight, delivers the next day

the sales cycle is almost too simple:

find a firm scared of cloud AI, bring a box to the meeting, set it on the table, let them touch the thing their data will live inside

close rate is 1 in 3 with zero marketing. every client came from the last one’s referral

the factory keeps shipping $600 units

he keeps turning them into $50k contracts

they’re in the hardware business

he’s in the trust business

[KingWilliam @kingwilliam_]

Anthropic just dropped the official playbook for running a company on Claude Code - 30 minutes, free, from the engineers who built it

the setup is one human as CEO and AI agents handling operations. the zero-headcount company stopped being a joke.

but watch what actually makes those agents reliable - they already know the job before anyone asks

here's the quiet part:
> every role is written down once, in plain instructions Claude reads on its own
> the agent loads it and runs without being re-briefed each session
> you review what it shipped, you don't re-explain the work

that's a skill. one file, written once, firing in every session forever.

and it's the exact piece people skip while they retype who they are into a blank box every morning

watch the playbook for the why. the 23 skills that turn Claude into that kind of teammate are in the article below

[Fokki @0x_fokki]

2023: AI could not feed Will Smith a noodle

2026: AI runs a six-tool animation factory

the gap took three years

one person ran them last month and made $12,345

six tools run the whole thing for $124/month: Claude, Midjourney, Runway, ElevenLabs, Suno, Make

most people are still laughing at the spaghetti

the build is in the article above

[Anatoli Kopadze @AnatoliKopadze]

instead of an hour of Netflix tonight, watch the man who quit Meta to bet against everything Silicon Valley is building

it's the clearest explanation I've seen of why today's AI is nowhere near as smart as everyone thinks, and what actually comes next

useful whether you've never touched AI in your life or have been using it every day for the past year

I turned the key ideas into a practical guide on getting 100% out of Claude

you can find it below

[Asteri @Asteri_eth]

ONE AI AGENT CAN TURN EMPTY BACKYARDS INTO POOL CONTRACTOR LEADS BEFORE THE HOMEOWNER EVEN ASKS FOR A QUOTE

It works like this: AI scans expensive homes, finds large backyards without pools, generates a realistic pool preview, estimates the potential value increase, and turns it into a personalized sales pitch.

It’s a fully automated workflow that:

> Finds homes worth $600K-$2M
> Detects large unused backyards with no pool
> Generates a realistic before/after pool visualization
> Estimates project cost and potential home value upside
> Turns everything into a ready-to-send offer for the homeowner or contractor

The valuable part is not that AI can make a pretty image.

It’s that the buyer sees the outcome before anyone sells it to them.

That changes the whole offer.

You are not saying "we build pools."

You are showing the exact pool that could exist in their backyard.

And in the article below you’ll find 4 more AI automations that can turn boring workflows into real revenue

[Charly Wargnier @DataChaz]

🚨 @Karpathy predicted the power of the "LLM Wiki." Google just formalized it.

Meet Open Knowledge Format (OKF): a vendor-neutral standard for giving foundation models the curated context they need.

I can genuinely see this replacing Notion, Obsidian, or traditional wikis for developer teams, and the reason comes down to bookkeeping.

Traditional wikis fail because humans inevitably abandon the tedious work of updating them.

As Andrej Karpathy pointed out recently, LLMs don't get bored.

They don't forget to update a cross-reference, and they can touch 15 files in a single pass.

OKF standardizes the interoperability layer so agents can actually do that heavy lifting autonomously.

Because the format is minimally opinionated, it doesn't dictate what you write, it just dictates how it's structured. You get:
→ Human-readable documents that live right alongside your code in version control
→ Cross-links that map out complex entity relationships without needing a graph database
→ A system that survives moving between different tools and organizations

There is no complex compression scheme.

No central registry.

If you can cat a file, you can read it.

If you can git clone a repo, you can deploy it.

This is how we stop rebuilding context pipelines from scratch every time a new model drops.

Announcement + spec file in 🧵↓

[Kshitij Mishra | AI & Tech @DAIEvolutionHub]

Someone on GitHub just shared a massive list of free projects that are ridiculously good.

Many of them can already replace software people are paying monthly subscriptions for.

1. TradingAgents
AI-powered multi-agent quantitative trading framework
https://t.co/xunNFf8ZxT

2. LibreChat
A unified interface for ChatGPT, Claude, Gemini, and other AI models
https://t.co/MJbeSPkVA1

3. HyperFrames
HeyGen's open-source video generation engine
https://t.co/qLPxS9JJL9

4. Fincept Terminal
An open-source alternative to the Bloomberg Terminal
https://t.co/CZvl5bAd63

5. MoneyPrinterTurbo
AI that generates short-form videos with a single click
https://t.co/7Bgikd2DsQ

6. Agentic Inbox
Cloudflare's open-source AI email assistant
https://t.co/LLRoslkkCL

7. VoxCPM
AI voice cloning platform
https://t.co/UzExPt8eFd

8. Flowsint
Open-source OSINT intelligence analysis tool
https://t.co/kBlOzgRKAs

9. agent-skills
A library of coding skills for Claude
https://t.co/Pxwwo42xnN

10. Nango
Open-source API integration platform
https://t.co/hQNCgSJtlg

These aren't toy projects.

A lot of the software you're still paying monthly for already has open-source alternatives built by developers on GitHub.

Some of the most powerful tools on the internet aren't being advertised.

They're hidden in GitHub repositories.

[Moon Dev @MoonDevOnYT]

retail traders get liquidated every minute while bots spot the data and instantly counter trade them on polymarket

this statistical arbitrage bot tracks massive liquidations on hyperliquid and hedges the other side for free money

stop staring at moving averages and watch how automated scripts exploit forced leverage liquidations 24 7

[Matt Dancho (Business Science) @mdancho84]

RIP MIT.

Top AI and ML repositories that replace a $250,000 college education.

(bookmark this for later)

[Movez @0xMovez]

Claude Code creator:

"100% of our pull requests at Anrtopic are run by Claude Code. 80–90% of code review too.

The feature I’m using the most today is /loops. I’m not prompting Claude anymore - I’m building loops"

in 1-hour interview, Boris reveals his setup, which helps him build the #1 coding tool of this year.

Worth more than a $500 vibe-coding course.

[Supermicro @Supermicro]

Want to see AI Factories in action?
Watch ServeTheHome’s latest video breakdown to see how Supermicro and NVIDIA deliver turnkey AI infrastructure at scale.

[Kawsar @Kawsar_Ai]

STOP TELLING CHATGPT “CHECK MY GRAMMAR AND WRITING.

Bad prompt = Bad result.

Use these prompts and you’ll see the difference:

[Vaidehi @Ai_Vaidehi]

Bro created an AI job search system for Claude Code that scored 700+ job applications and actually got him a job.

AND IT'S NOW OPEN-SOURCE.

It scans multiple company career pages, rewrites your CV per job, and even fills application forms. The repo has:
> 14 skill modes (evaluate, scan, PDF, ...)
> Go terminal dashboard
> ATS-optimized PDF generation via Playwright
> 45+ companies pre-configured (Anthropic, OpenAI, ElevenLabs, Stripe...)

Giving This Free for 24 hours. To get it:

1. Comment the word 'NEED'
2. Like and Retweet this post
3. Follow me @Ai_Vaidehi
(so i can DM you)

[Arsh Shah Dilbagi @arshdilbagi]

Introducing Adaline 2.0 - The Agent Self-Improvement Layer

Adaline turns Traces into Behaviors,
Behaviors surface Issues,
Issues become auto-generated Evals + Data,
Adaline then generates new agent candidates and tests them.

You review the winners and ship!

[Vaibhav Sisinty @VaibhavSisinty]

A 25-year-old housewife in Chennai earns ₹250/hour ($3) just by doing her normal housework.

She wears a phone on her head and records herself making coffee, cutting fruit, folding laundry.

These first-person videos get sent to AI companies training humanoid robots to handle real-world tasks. She shoots 90+ clips a day.

Her quote: "Who else will pay you ₹250/hour ($3) an hour just for doing housework?"

She's part of a growing gig economy in India where thousands are doing the same thing, filming everyday life to train the robots of tomorrow.

[Gipp 🦅 @gippp69]

I’M 18 AND MY $20 CLAUDE GAME IS STARTING TO LOOK LIKE A $3,000 BRAND DEMO I COULD PITCH TO A COMPANY

the first build was just cars, roads and basic movement. now it has missions, traffic lights, wider streets, named shops, motel signs, house labels and a cleaner city layout

I also upgraded the small details that make it feel less empty: NPC conversations, better textures, stronger lighting, smoother drone shots and police that actually step out of the car

this is the part where it stops feeling like a toy. once the world has goals, signs, cops, UI, traffic and people reacting around you, the demo starts selling the idea without a pitch deck

a normal agency would package this as an interactive campaign, add a brand skin, change the map, put a product in the mission loop and quote somewhere between $750 and $3,000

mine still runs as a simple browser game from a cheap Claude workflow, but every new feature makes it easier to imagine as a client demo, product launch game or viral playable ad

I’m still building it. comment what I should add next: interiors, better missions, smarter cops, shops, gangs, day-night cycle, car upgrades or anything that would make this more sellable

[Marie Haynes @Marie_Haynes]

This is really big news. Google introduced the Open Knowledge Format (OKF) - a standardized way to store information in a directory of markdown files. Makes it really easy to make a digital brain that agents can use.

These files can serve as a living wiki. You can give agents the ability to query them or edit them. They can interlink.

Seems to me this could replace Notion or Obsidian. I can think of so many uses for this.

Google's blog post: https://t.co/DqSjg4UpvH

An easier to understand explanation is the SPEC.md file:
https://t.co/A3qSz3Tfas

I gave those two links to Antigravity and asked how we could use it for any of the projects we're working on. It came up with so many ideas. I would imagine Claude Fable 5 would whip up some pretty amazing things based on this system.

Currently creating an OKF library of our pepper garden. It's going to be a fun weekend.