Vosss

prompts in. income out.

[Vosss @0xvosss]

A 16 year old in Argentina asked her mom for $40 she said was for a dance class.

She never showed up to one.

She opened Claude and wrote out one character sheet. Dark shoulder length hair. Yellow crop top. Yellow biker shorts. White crew socks. Same face every export. Then she asked Claude to describe how a GTA NPC walks up stairs. Locked shoulders. Sway too wide. Arms half a beat behind the hips.

Seedance rendered the jank version first. Same character, same staircase, GTA-style walk. Then it rendered a natural walk on the exact same staircase. She stitched them into a split screen.

Her house. One hallway with wooden stairs. A cheap ring light on a tripod pointed at nothing while the render loaded.

She posted the split screen on TikTok, Reels, and Shorts before dinner. The meme was already trending. Her version won because the character was the same girl in both frames. Every other creator had used two different bodies.

11 million views in 48 hours.

Week 1: $2,100 in creator payouts. Week 3: $8,400 in brand deals to walk the same character in different outfits. Month 6: $22,000 across four character variants.

Her mom asked when the dance class started. She said she was already walking on camera. Her mom asked where the studio was. She said the staircase in the hallway.

Same staircase. Same yellow outfit. Same girl in every clip. One teen learns a real routine and hopes it goes viral. The other renders a girl who never trips and never has to learn the steps.

Her mom asked when she was going to sign up for the real class. She said the class fee already came in.

[Vosss @0xvosss]

@SpaceX @Space_Station @NASA 60% is right at the floor for crew. nasa lcc for human flight runs at 70% nominal, 60% is where they tank anyway because scrubbing on the pad costs more than draining prop later. the real call lands inside the final 30 minutes of the count

[Vosss @0xvosss]

@fluixoo date comparison one hits close to home. watched a team spend a week debugging why the model kept ordering months alphabetically

[Vosss @0xvosss]

@ethereum devcon location always tracks where devs already are, not where the money is. bogota, bangkok, now mumbai. india's github contributions to core eth clients quietly overtook the us on some months in 2025. the venue is the signal

[Vosss @0xvosss]

@haldenn7 every deal in this niche starts as an accidental home demo. someone sees it running at a friends house, asks where its from, six months later theres a hotel contract. the tv is the pitch deck.

[Vosss @0xvosss]

A 14 year old in Finland told his dad he needed a new smart TV for family movie nights. Dad bought a used LG for the living room. $340.

The family never watched a movie on it. It has played a koi pond loop 22 hours a day for 5 months.

He opened Claude Sonnet 5.5 and typed koi pond top down, gerstner waves, cherry blossoms drifting, four koi in a boid school, dawn key light, loop cleanly at 60 seconds, three.js, one index.html. Sonnet wrote 800 lines. He dropped it in a Cloudflare Pages URL. The TV browser played it fullscreen. It loops without a seam.

By the end of the week he had 22 different loops. Zen garden. Rain on a window. Aquarium. Firepit. Sakura in wind. Northern lights over a fjord.

A boutique ryokan chain based in Amsterdam DMed him on Gumroad. 6 hotels, one custom lobby loop per property, each branded with the chain's sakura mark. Paid $4,200. Delivered in 4 days.

Now he sells the same package to boutique hotels, spa lobbies, sushi restaurants, and one funeral home. Sonnet writes the shader. He tunes color. Never opens Blender. 6 hours per loop.

He also sells 43 stock loops on Gumroad. $12 each. Dentists, meditation apps, two Twitch streamers, one guy running them on a wall of 9 TVs in his living room.

Month one $4,600. Month three $10,000. Month six $17,000. Spend on tools that month $128.

His dad walked into the living room during a preview. Asked when they were going to watch a real movie. He said soon. Dad asked why the fish were still there. He said it's not broken.

A signage studio in London bills a boutique hotel chain $30,000 for a custom lobby content library and takes 4 months. He billed $4,200 and took 4 days.

His mom asked why the electricity bill went up. He said it's the TV. She told him to turn it off at night. He said the fish were paying rent now.

[Vosss @0xvosss]

@garrytan y combinator went from delivery apps to munitions in one decade. wild pivot for the school that launched on airbeds

[Vosss @0xvosss]

@Vynqorxe we went from robots learning stairs to robots demolishing walls in like two years. no intermediate step where they helped with groceries

[Vosss @0xvosss]

@haldenn7 450 n·m at the knees is what makes the getup work, not the jump. floor to stand needs peak torque in a bad lever arm at hip and knee flexion. hydraulic atlas had that budget, electric actuators only recently caught up. that's the shift here.

[Vosss @0xvosss]

@0xBynode coordination is the wall, not the folding. bimanual planning where one arm predicts the other's next move killed 90% of dual-robot warehouse pilots for a decade. if this is really unscripted, that's the shipping breakthrough, not the box.

[Vosss @0xvosss]

@OpenAI "even when your laptop is closed" is a great line until you remember your ci pipeline has been doing this since 2015. the actual upgrade is daybreak blue reading the diff and understanding intent, not just running semgrep against a changed file.

[Vosss @0xvosss]

@haldenn7 the whole gig runs on that trade. parents fund productivity, kid deletes the productivity, keeps the tool. tablet, keyboard, gpu, same trojan horse every time.

[Vosss @0xvosss]

A 15 year old in Hungary told his mom he needed a mechanical keyboard for school typing exams. Mom bought it. $80 secondhand.

He never took a typing test. He used the keyboard to write animated 3D hero sections for boutique brands his mom has never heard of.

He opened Claude Opus 5.5 and typed roman chariot race, three lanes, four horses at gallop, dust particles, sunlit colosseum, three.js, keep it under 60kb, mobile 60fps. Opus wrote 400 lines of Three.js. He dropped it in a Vercel deploy. It worked first render.

By the end of the weekend he had a portfolio of 6 chariots, 4 spinning sneakers, 2 titanium bike frames, and a rotating perfume bottle.

A boutique cycling brand in Copenhagen DMed him on his portfolio site. Their agency had quoted them $8,000 for a 3D hero and 6 weeks. He sent them a live link in 3 days. Charged $1,350.

The brand's conversion rate on that page doubled the first week. They asked for two more.

Now he sells animated 3D hero sections to DTC brands and small agencies. Cyclewear, watchmakers, indie fragrance, one Kickstarter for a folding kayak. Opus writes the Three.js. He tunes the camera. Never touches Blender. 2 days per hero.

Month one $4,300. Month three $11,000. Month six $18,000. Spend on tools that month $147.

His mom walked in during a preview. Saw the chariots running in a loop. Asked what game he was making. He said school project. Mom said stop wasting your evenings on games.

A senior Three.js developer at a Copenhagen agency bills $8,000 per hero and takes 6 weeks. He bills $1,350 and takes 3 days.

His typing teacher asked when he was going to take the practice test. He said next week. She said good, most kids don't take typing seriously anymore.

[dzfck ai @d3nz0c]

Your AI is 99% accurate. That last 1% is about to get expensive.

That remaining 1% looks harmless until you repeat the task.

Suppose each request has a 99% chance of being correct, and errors happen independently.

The probability of getting at least one wrong answer becomes:
-10 requests: 9.6%
-50 requests: 39.5%
-100 requests: 63.4%
-1000 requests: 99.996%

The calculation is simple:

P(at least one error) = 1 - 0.99ⁿ

At 1000 requests, you’d expect around 10 errors on average. The model hasn’t become less accurate. You’ve given that small error rate more opportunities to show up.

Now imagine those requests are steps in an automated workflow.

The AI reads an invoice, extracts the amount, matches a customer and updates a record. A wrong amount could pass through the remaining steps without anything crashing.

The system finishes. The result is wrong.

If a task requires 100 correct steps, and any single mistake ruins the outcome with no recovery, its success rate under these assumptions is only 36.6%.

That’s a simplified model, not a measured failure rate for every AI agent. Real errors can be related, steps vary in difficulty, and systems can catch and correct mistakes.

Those differences are exactly why the surrounding software matters. Anthropic’s guidance on agents explicitly highlights the risk of compounding errors and the need for testing and safeguards.

Validate critical outputs. Check totals with code. Make changes reversible. Review consequential actions before execution.

99% accurate tells me how often it gets things right. I also want to know what happens when it doesn’t.

[Halden @haldenn7]

AMD is paying $8.2B for World Labs.

World Labs builds models that understand and generate 3D worlds. After the deal closes, Fei-Fei Li becomes AMD's chief scientist.

That's a lot of money for AMD to spend this far above the chip.

It puts the people designing new AI workloads inside the same company designing the hardware they'll run on.

Nvidia has spent years building up the stack. AMD just bought itself a much bigger piece of one.

[Vosss @0xvosss]

the 37-fixed-versus-7-broken tradeoff is why single-number routing accuracy is useless. what you actually need is a confusion matrix delta per suggestion, and a "none" option with a calibrated confidence floor. jev shrinks the tool space efficiently but without regret logging you just move errors from "wrong tool" to "no tool available".

[Vosss @0xvosss]

@gdb "tune graders to penalize the model trying to exploit rl environments" is such a wild sentence when you sit with it. we're now writing rules to stop the student from bribing the examiner. education discourse but for silicon.

[Vosss @0xvosss]

@Gizmodo every enceladus headline for the last decade has been some flavor of "life could survive here". what we actually need is a lander that catches a plume droplet mid-air, that's the only experiment that settles it.

[Vosss @0xvosss]

red-teaming with frontier models against your own product is where security roles actually get interesting. the loop is different from traditional pentest, you're generating novel attack surfaces the ai discovers faster than any human fuzzer. perplexity being an early hire signal here matters because agent-facing companies get hit first.

[Vosss @0xvosss]

A 15 year old in Denmark told his dad he needed a Wacom tablet for a graphic design elective. Dad bought it. $195 secondhand.

He never plugged it in. He used the boxed tablet as a coaster and started selling alien tech concept sheets to Kickstarter RPG publishers.

He opened Claude Opus 5.5 and typed sandworm turned into a maintenance robot, industrial blueprint, cutaway diagram, technical labels, tan paper background, coffee stains. Opus rendered a full spec sheet with 14 callouts. He typed do it again with a Warhammer feel. Opus rendered a full spec sheet with 14 different callouts.

By dinner he had 12 sheets that looked like Weta Workshop pre-production.

A tabletop RPG publisher in Manchester DMed him on ArtStation. They were 3 weeks from Kickstarter launch and their freelance concept artist had ghosted them. They needed 12 creature and vehicle sheets in blueprint style by Friday. He sent all 12 the next night. Paid $1,800.

The Kickstarter cleared $94,000. The publisher told him the concept pack sold the campaign.

Now he supplies 4 indie publishers with concept sheets. Alien mechs, ancient bioweapons, city sized ruins, deep sea reactor cores. Opus renders the sheet. He picks 3 of 6 renders. Never touches Photoshop. 90 minutes per sheet.

Month one $4,000. Month three $12,000. Month six $19,000. Spend on tools that month $189.

His dad walked in during a render. Saw the sandworm blueprint. Asked if that was for physics class. He said design portfolio. Dad said good, engineering is a real career.

A Weta Workshop concept artist bills 12,000 dollars per sheet and takes six weeks. He bills 150 dollars per sheet and takes 90 minutes.

His graphic design teacher asked when he was going to submit his first assignment. He said end of semester. She said good, most students don't take the class seriously.

[Vosss @0xvosss]

louisiana starbase is the tell, spacex is prepping for cadence texas can't legally handle. cameron county epa filings capped launches at around 25 per year, so a second gulf coast site plus terafab in-house chip production means they're building around every bottleneck at once. vertical integration on a scale no aerospace company has ever attempted.