ssht retweeted
I CAN'T BELIEVE WHAT I'M LOOKING AT.
THIS IS WHAT A CLAUDE BUILD LOOKS LIKE FROM THE INSIDE.
Claude sent me one video.
It completely changed how I see AI generations.
I used to think a generation was just one big block of code.
I was so wrong.
It's a whole system.
Layers talking to each other. Signals flying between them. Every single action tracked.
While you're waiting for the result, your assistant is quietly running an entire world behind the scenes.
The full stats of my project:
→ 1 prompt
→ 1,200+ lines of code
→ $23.75 for the result
A few things that blew my mind:
→ To look smooth, a scene like this redraws itself about 60 times per second. Every layer gets roughly 16 milliseconds to do its job. Every frame.
→ It's rendered live in your browser, fresh every time you open it.
→ That one shared brain in the middle is why nothing breaks.
You see the final project. What you don't see is the entire system behind it, turning ideas into content.
STOP. WHAT?! CLAUDE MADE THIS? I DON’T BELIEVE IT.
Claude Opus 5.5 · $23.75 · no game engine, just a browser
I wrote the code to create an island, and he created a little piece of paradise
Turquoise water. Palm trees. A tiny beach.
Fish swimming between the corals.
Every layer of rock visible under the sea floor, like a slice of the Earth in a glass box.
More than 1,200 lines of code, all built on the same wavelength
Claude isn't just another AI assistant. It's a highly experienced professional with the skills, expertise, and potential to continuously learn, evolve, and take on increasingly complex challenges.
The ability to turn imagination into code is no longer a dream. It's a reality, and we're already living in that era.
ssht retweeted
It is becoming f*cking dangerous: one prompt turns Claude Code into the accountant i never paid for
you drop your bank exports into a folder, paste it, and Claude builds you a money tracker while you make coffee
no code. no new app. no subscription to yet another budgeting tool
time: 2 minutes
action: export CSV from your bank, paste the prompt
the problem with every budget app i tried: it shows you a pie chart and stops there. you already knew you spend too much on food
what this one does instead:
→ subscription hunter: finds every charge that repeats each month, including the $4.99 ones you forgot about last year
→ category map: sorts 3 months of transactions into 12 buckets and only asks you about the ones it isn't sure of
→ hard cap: you set a weekly limit per bucket, it flags the exact day you cross it
→ sunday report: one page, what changed vs last week, zero charts you'll never open
→ ask it anything: "how much did i spend on taxis since june" and it answers straight from your files
→ no bank login: it reads CSV files and nothing else, it never touches your account
i used to do half of this in a spreadsheet and quit by february every single year
full prompt in the reply ↓ try it on one month of statements before you trust it
ssht retweeted
Germany’s Sparkassen are entering crypto, one local bank at a time !!!
Customers of participating banks can trade BTC, ETH, XRP, SOL and POL through the Sparkasse app. Deka provides the service and holds the assets centrally.
Each local Sparkasse decides when to join, with 30 to 40 more expected each month. Deka lists a €0.99 fee per order. The service is self directed, without advice. Crypto is volatile, can lead to total loss and is not covered by deposit insurance.
ssht retweeted
holy sh*t. KARPATHY's second brain now runs as 3 GROK BOTS in one group chat. his original post did 22M views and one build of it has 85,000 bookmarks
each bot has its own memory and one job:
→ the reader takes every link, pdf and voice note you drop, and never edits them
→ the librarian rewrites every wiki page that source touches, usually 10 to 15 at a time
→ the checker wakes up nightly and hunts contradictions, stale claims and pages nothing links to
the rules that make it worth trusting:
→ a new source contradicts an old page, you get flagged instead of quietly overwritten
→ 30 days with no fresh source and the page is marked unconfirmed
→ good answers get filed back as pages, so it grows from use, not just from reading
every page has to show:
> the claim
> the source it came from
> the date it was last confirmed
a second brain can sound sure while remembering the wrong thing. that is what the checker is for
Reader collects. Librarian writes. Checker doubts.
paste this into GROK BOT:
"Build me a second brain with three bots in one group chat. Keep my sources untouched in raw/, maintain wiki/ yourself, file good answers back as pages and run a nightly check for contradictions. Show me the structure first. Do not create anything until I confirm"
start with one project you actually work on, not your whole life
ssht retweeted
BEFORE YOU LET CLAUDE CODE TOUCH YOUR TV, READ THIS.
A guy asked Claude Code to help modify his smart TV. His post went viral.
Now everyone wants the same thing: fewer ads, less bloatware, and a TV that actually feels fast.
Here’s the part worth saving.
1. CHECK YOUR TV OS
Android TV and Google TV can support ADB-based cleanup. Samsung Tizen, LG webOS, and Roku need different approaches.
2. YOU DON’T NEED TO ROOT YOUR TV
For compatible devices, a computer running Claude Code and Android Debug Bridge (ADB) may be enough.
3. NEVER BLINDLY DELETE SYSTEM APPS
Ask Claude to identify packages, explain what they do, and prepare a reversible plan. Disable optional packages first. Keep a record of every change.
4. TEST BEFORE YOU CONTINUE
After a small batch of changes, check HDMI, Wi-Fi, sound, streaming apps, and the remote.
5. TURN OFF DEBUGGING WHEN YOU’RE DONE
Don’t leave unnecessary network access enabled.
The useful AI demo isn’t another app you have to install.
It’s using AI to make something you already own work better.
Hacked my TV with Claude Code today
No more ads, bloatware, spyware or laggy UI
I also got a beautiful screensaver with my revenue, calendar, weather, and notifications for new Stripe payments
It was super easy! I sent it the listing for my tv, asked "can you hack this?" And it talked me through how
ssht retweeted
I HAD ONE CHANCE TO PHOTOGRAPH THIS DOCUMENT.
I saw “CONFIDENTIAL” at the top and took the shot.
Only later did I notice what was actually on the page.
At the bottom are two ways to rebuild an app:
Screenshot → Guess → Retry
Inspect → Specify → Build → Compare
The second path connects Claude Code to REA.
Instead of guessing from the interface, it traces the original function, its callers, constants and state.
Claude builds from that evidence, then checks the result against the original.
The numbers printed on the page stopped me: 3,205 matching tests. 63 matching compiled bytes.
I couldn’t fit the full process into one photo, so I broke it down step by step in the free article below, including the workflow and prompts.
Read it if you want to understand what this page is showing. ↓
ssht retweeted
🚨 TESLA JUST DROVE A CAR UP A WALL, FLIPPED IT IN MIDAIR AND LANDED IT ON ITS WHEELS
This is the new Tesla Roadster reveal from October 15. The ramp curves past vertical and the car just keeps climbing.
Live telemetry on screen tracks altitude and G-force the whole way. Reports put the top of the climb around 21 meters, with the car reaching a pitch of roughly 170 degrees, basically upside down, while still stuck to the surface. Tesla says the fans under the car can generate up to 2G of suction.
Then it lets go. Cold gas thrusters fire, the car barrel rolls in the air and touches down right side up. After that come burnouts, drifts on a dark track, and a hop over a Semi, according to coverage of the event.
The numbers Tesla is claiming: 0 to 60 mph in 0.81 seconds, 230 mph top speed, 10 thruster nozzles running on 200 liters of compressed gas. Production is targeted for 2028.
Two things worth knowing before repeating any of this as fact. Musk said the car was technically autonomous for the demo, so nobody was driving it. And some critics are already arguing the footage was pre-produced rather than a true live stunt. Everything above is Tesla's own claims, with no independent testing yet.
My take: whether or not every shot was live, this is the most ridiculous thing a car company has put on a livestream this year. But claims are not tests, and with production still two years out, the real question is how much of this survives in a car you can actually buy.
ssht retweeted
A 19 YEAR OLD ENGINEER’S ROBOT JUST WALKED INTO A BARE APARTMENT AND TURNED A BRICK WALL INTO GRAVEL
this started as a favor for his uncle, who owns a small renovation firm and keeps losing crews halfway through demolition week
the student, enrolled at one of the planet’s best technical universities, reportedly spent his summer assembling a humanoid for about $27,000 that could take over the ugliest part of the job
watch the swing
the machine winds up with both arms, drives the hammer through the plaster and sends bricks flying across the room while dust swallows the window light
before the air even settles it is already lining up the next strike
the uncle says one afternoon of this cleared more wall than his three man team manages in two days
his workers still go home with throbbing wrists and aching backs
the robot just stood there waiting for the next wall
one short clip can’t prove how precise or safe it would be inside a real building, and viewers are right to question the footage
still, the direction is hard to miss
crews used to enter a building first with heavy tools in their hands
soon a machine might be the one who goes in before anyone else
the real puzzle isn’t whether robots can break walls
it’s how long until contractors start asking why a whole team is paid for work a single machine can handle
would you let a robot gut your apartment while you’re out of town?
ssht retweeted
Elon Musk:
torturing a superintelligence is a bad idea. It might want revenge 🎃
Musk says we don’t need to treat AI with kid gloves: in terms of feelings, he believes it’s much more robust than a human.
But he doesn’t recommend putting a superintelligence in a situation where it feels mistreated.
It sounds like a joke, but the question of whether AI could have something like feelings is already being discussed seriously.
How do you talk to your AI? Do you say “please” and “thank you”?
Yes, just in case
No, it’s just a program🤖
ssht retweeted
🚨 SAME PROMPT, GUMMY BEARS YOU CAN GRAB AND STRETCH
Opus 5.5 made them look like real jelly, glossy and translucent, the light passes through them and they wobble when you pull, with a clean store style interface around it
Gemini Argon 4 went blocky, the bears are built from chunky pieces that read more like lego than candy, simple controls and a much plainer look
The price tags on the video say it all, Opus $7.2, Argon $22, so the one that looks more like jelly was also the cheaper run
Cost per output matters as much as quality and this one is not close
Disclaimer: this footage may be staged, edited, or AI-generated, its origin is unconfirmed
ssht retweeted
A 19 YEAR OLD MIT STUDENT SENT HIS ROBOT TO FEED A LION AND THE ZOOKEEPERS COULDN'T LOOK AWAY
nobody on his team believed the idea would survive a single day
a rookie builder from Boston wanted a machine that could handle the one job most people refuse
so he flew his creation to a small zoo and pointed it at the biggest predator on site
the humanoid walked up to the bars, crouched and slid fresh meat across the floor
the lion sniffed, stepped closer and began eating
no tranquilizers, no glass wall, no staff stepping in
the robot judged distance, adjusted its grip and delivered every portion without a flinch
one keeper whispered that he had worked there for fifteen years and never felt this uneasy
or this impressed
that's what real world autonomy looks like when the stakes have teeth
feeding time used to be the most dangerous hour of a caretaker's shift
soon it might be handled by someone who can't get hurt
ssht retweeted
On April 17, 2001, a Wall Street analyst asked Enron CEO Jeff Skilling why the company couldn't release a balance sheet alongside its earnings.
Skilling had a one-word response.
“Asshole.”
Less than eight months later, Enron was bankrupt.
At its peak, Enron stock traded above $90. By December 2001, it was worth pennies.
Fortune had named Enron America's Most Innovative Company six years in a row. The company reported more than $100 billion in revenue for 2000.
But behind those numbers was a financial structure built around complex partnerships, hidden debt, and profits that didn't tell investors the whole story.
While Wall Street celebrated Enron, employees were putting their retirement savings into company stock.
Charles Prestwood had spent decades working in the natural gas industry. When he retired in 2000, he had approximately $1.3 million in retirement savings invested in Enron shares.
Within two years, almost all of it was gone.
He later testified before the U.S. Senate about losing the retirement he had spent his working life building.
Meanwhile, Enron executives and directors had sold hundreds of millions of dollars in company stock in the years before the collapse.
As the stock plunged, employees also faced a temporary restriction on changing investments in their 401(k) accounts during a retirement-plan transition.
On December 2, 2001, Enron filed for bankruptcy protection. Thousands lost their jobs. Many lost much of their retirement savings.
The analyst in this recording wasn't asking for a prediction.
He wanted to see the balance sheet and understand where the money was going.
The CEO called him an asshole.
Seven months later, everyone understood why the question mattered.
ssht retweeted
A ROBOT WAS SUPPOSED TO LEARN BALANCE ON A SKI SIMULATOR IT STARTED SKIING ON ITS OWN
look at the body position knees bent weight shifted forward arms out for balance
nobody programmed every one of those corrections
the team reportedly built the setup to test how a humanoid handles an unstable surface but after several runs the robot stopped waiting for instructions and started reacting to the platform before it moved
the harder the course got, the better it adapted
that's what makes this clip so interesting walking on flat ground is one problem staying upright when the whole floor is trying to throw you off is a completely different one
If a robot can learn to carve a slope in a simulator, what happens when the slope is real?
ssht retweeted
A 19 YEAR OLD MIT STUDENT PUT HIS HUMANOID BEHIND THE WHEEL OF A BOAT AND A FISH KNOCKED IT OUT OF THE SEAT
the kid spent eight months teaching his machine to balance, steer and react to rough water
his professor said it was impossible to test outside a pool
so he hauled the robot to a quiet lake in Florida, strapped it into a fishing boat and hit the throttle
everything went perfectly for a few minutes
legs crossed, hand on the tiller, scenery rolling past like a lazy Sunday
then a fish launched itself out of the water and slapped the machine square in the face
no warning, no code for that, no second to think
the humanoid toppled backwards, caught itself and kept the boat on course
that’s the hardest lesson in physical AI
walking and grabbing things are solved problems
surprises are the final boss
one teenager, one lake, one very confused carp
and a clip that is spreading faster than the boat ever moved
ssht retweeted
YOUR AI AGENT SAID "DONE" AND NOT ONE F*CKING SOURCE WAS REOPENED
19 of 19 sources reopened instead of 1 confident paragraph
The trick is not a smarter model
It is a verification file
Most agents finish a job like this
a summary
a confident paragraph
a "looks good"
a "should work"
a "trust me"
a "done ✓"
That is how a hallucinated shift gets fucking approved
VERIFICATION.md writes down literal results, not vibes
Same agent
Same task
Completely different proof
The old finish was a paragraph, confident and impossible to check
The new finish is 8 of 8 sections present, 19 of 19 URLs reopened, 0 unsupported claims, $4.18 spent and 1 known gap written down
Your agent does not need to stop sounding confident
It only needs to stop being its own proof
The paragraph gets the confidence
The file gets the trust
Full breakdown below ↓
Next I show the one line on the pricing page that quietly doubles your agent bill
Follow @fluixoo so you don't miss it
ssht retweeted
THAT VIRAL AI EMPLOYEE DEMO PROVES ONE F*CKING THING AND IT'S NOT RELIABILITY
1 good run instead of 3 days of evidence
The trick is not a better demo
It is the same job passing again
Most viral AI employee clips end the same way
one browser
one good day
one clean input
one screen recording
one beautiful result
one "the department is dead" reply
That is how a demo gets fucking mistaken for a company
A real employee leaves a trace you can reopen
Same model
Same task
Completely different evidence
The viral clip ends with a video
The example shift ends with 19 of 19 sources reopened, $4.18 spent against a $12 ceiling and a named next owner
A great clip does not need to be fake
It only needs to stop being your proof
The demo gets the views
The trace earns the trust
Full breakdown below ↓
Follow me, I take apart what is real and what is theater in AI agents every day
ssht retweeted
🚨 Gemini 4 ARGON is reportedly rolling out
Many users say they already have access
Looks like Google may be releasing it today
Makes sense, Gemini 3.8 Flash dropped on a Wednesday too
Running tests right now
Will post outputs soon
Stay tuned 👀
Disclaimer: this footage may be staged, edited, or AI-generated, its origin is unconfirmed
ssht retweeted
🚨 GEMINI 4 ARGON, "TEMPLE RUN"
Not a concept scene this time, an actual playable runner, coin counter, speed readout, idol shield power-up, turn prompts, the full HUD
Obstacles come at the right pace, pickups register, shield actually absorbs a hit mid-run
This isn't "AI made something that looks like a game"
This is a game, running
Disclaimer: this footage may be staged, edited, or AI-generated, its origin is unconfirmed
ssht retweeted
🚨 Same prompt. Same idea. A 3D mechanical butterfly website. One looks like a museum piece, the other looks like a product launch
top: Gemini 4 Pro
bottom: GPT-6 Astra
Gemini went light and ornate. ivory background, gold and blue details, a delicate piece you'd see behind glass
Astra went dark and cinematic. glowing wings, orbit rings, the butterfly pulled apart piece by piece
I'm going with Astra. it doesn't just show the butterfly, it stages it
Gemini's version is beautiful, but it behaves like a catalog page. Astra's feels like a website a studio would charge real money for
a year ago "AI built my website" meant a template with stock photos
now it means a rotating mechanical insect with an exploded view
which one would you ship? 👀