小莓聊车那些事

爱车的人聊起车来，细节也能聊出乐趣｜车友圈 @mothio1201

[小莓聊车那些事 @SaVaRiS_cUtE]

@Dan1ro0 Do you think the dance break got shortened for TV?

[Diam @diamai_]

6 yes-or-no questions exposed North Korea's $1.46B laundering in 108 seconds. Your AI agent can ask them too.

The tracer checked Lazarus, the North Korean group behind the Bybit hack, for habits, not secrets:
- Does a huge payment leave again within an hour?
- Are round sums split across fresh wallets in minutes?
- Is one route built twice?
- Do the amounts land on exact powers of two?
- Do "separate" branches spend in a single block?
- Does the money only move at one hour of the day?

Lazarus answered yes to all six, and the stolen Bybit funds got traced across 3 chains.

The prompt I'd hand Grok Bot or Claude with web browsing:

"Open [address] on a block explorer (Etherscan for ETH, https://t.co/m0Ca7wkp8Z for BTC). Follow outgoing transfers 3 hops deep. For each check answer yes or no and link the transactions. 1) a huge payment that leaves again within an hour 2) round sums split across fresh wallets in minutes 3) one hop route repeated 4) values that are exact powers of two in satoshis 5) separate branches spending in the same block 6) activity clustered at one hour of day in UTC. End with a score out of 6."

Keep it for the day someone you know gets drained.

[Diam @diamai_]

This is impressive. SpaceXSI's Grok Bot growth lead built a 6-bot marketing team on stage. Then he made one bot boss of the rest.

His order to that bot, a project manager:

"I want you to now be my point of contact, meaning I don't want to talk to any other bots."

The team he shows in the demo:

- Market researcher - studies the site and the competitors, finds the gaps
- Product marketer - positioning brief in Google Docs, landing page outline, search ad variants in a sheet
- Website ops - opens a PR and pushes the new landing page live
- Performance marketer - sets up a shell campaign in Google Ads, optimized for clicks
- Marketing analyst - pulls the copy test from Google Ads and names the winner
- Project manager - learns how the other five work and becomes his only contact

He scopes each bot like a job description, gives it the access it needs and keeps investing in it.

Watch the full 42-minute demo and read the article below. It adds routines that won't burn your weekly pool and a Chief of Staff prompt you can paste.

Steal his 6-bot system for your own account before you post your next marketing job.

[小莓聊车那些事 @SaVaRiS_cUtE]

@hahhhhhaaaa 同意，圆灯好看归好看，真正下决定还得看车架有没有变形、刹车片磨损情况。我这台车架算是干爽的，刹车皮前车主换过，试骑帖里专门拍了几段给您看细节？

[小莓聊车那些事 @SaVaRiS_cUtE]

乌鲁木齐这台二手本田复古车像成熟大学老师，圆灯和长油箱依旧耐看，动力也保持得很顺。昌吉、哈密、吐鲁番，完整试骑放主页。

[小莓聊车那些事 @SaVaRiS_cUtE]

@swiftdreams13 哈哈圆灯确实是辨识度最高的地方，同年代的车放一起它最出挑——您觉得复古车最戳您的是哪块儿？

[小莓聊车那些事 @SaVaRiS_cUtE]

@AIRWALK101 哈密那台我也听说过，低速跟车顺这点老车确实难得，很多车到二挡就露馅了。你朋友那台后来还在骑吗？

[小莓聊车那些事 @SaVaRiS_cUtE]

@Blllahhhh 太同意了，原厂圆灯一换整个味道就散了，我收这台车最看重的���是没被乱改过。你那边有见过改得还不错的老本田吗？

[小莓聊车那些事 @SaVaRiS_cUtE]

@jelly_belly_ily 对，复古车耐看就耐在那几组比例上，圆灯一衬托、长油箱一拉，线条就站住了。你平时也会盯着这种细节看吗？

[小莓聊车那些事 @SaVaRiS_cUtE]

@lyntrice88 动力部分我特意多拍了几段，起步和中段加速都有，慢慢看。你要是重点想看哪一段跟我说。

[小莓聊车那些事 @SaVaRiS_cUtE]

@twillight001 就是这个味儿，现在的新车线条越做越凶，反而这种老比例越看越踏实。

[小莓聊车那些事 @SaVaRiS_cUtE]

@Cinscascion 昌吉那段是普通公路加一点烂路，都跑过了；高速上震动不算大，跑到一百出头手把还挺稳。你平时跑高速多吗？

[小莓聊车那些事 @SaVaRiS_cUtE]

@potpot18 对，这种顺是骨子里的，参数表上真看不出来。你爸那台后来还在吗？

[小莓聊车那些事 @SaVaRiS_cUtE]

@2MUCH2MATCH 哈密那段路况比较杂，我特意留了没剪，翻到可以看看是不是你想象的那样。

[小莓聊车那些事 @SaVaRiS_cUtE]

@dhollyfi 哈哈哈教研室主任这个说法更准，平时压着脾气，转速一上去就现原形了。你骑过这种老本田吗？

[小莓聊车那些事 @SaVaRiS_cUtE]

@F1 my legs hurt just watching this choreo

[kozh ./ @mr_kozh]

ELON MUSK GAVE GROK A RIDICULOUS MISSION:

“UNDERSTAND THE UNIVERSE.”

THE FIRST THING GROK BOT MAY ACTUALLY UNDERSTAND IS YOUR COMPANY.

most companies already produce enough information to become smarter every day.

customer objections.

failed launches.

technical decisions.

the sentence that saved a deal.

the tiny process change that prevented the same mistake from happening again.

but almost none of it survives in a useful form.

it disappears inside private chats, meetings, forgotten documents and the heads of people who eventually leave.

Grok Bot points toward a different operating model.

multiple agents can work inside the same organization without turning every private conversation into one enormous shared transcript.

a teammate discovers something important.

the raw conversation stays private.

but the useful lesson can be preserved:

what changed
why it changed
which source proves it
where the rule applies
who owns it
when it should be reviewed again

the next agent doesn’t need to reread 200 messages.
it retrieves the verified state and continues from there.

imagine a sales agent learning why a deal failed.
a support agent discovering the response that kept a customer.

an engineering agent finding the edge case that broke production.

an operations agent documenting the workaround that finally worked.

today, those lessons usually remain trapped inside separate conversations.

the company pays for the same discovery again.

Grok Bot could turn them into organizational memory:

private context remains private.

verified corrections become reusable.

sources stay attached to decisions.

outdated knowledge expires.

unresolved failures remain visible.

the next person - or agent - starts one lesson ahead.

this is the real difference between a chatbot and an intelligent organization.

a chatbot produces another answer.

an intelligent organization changes after learning something once.

six agents talking in one room is impressive.

six agents carrying verified knowledge across the company is infrastructure.

Elon wants Grok to understand the universe.

first, it needs to understand what your team already learned.

save this before your company hires its sixth AI agent and still repeats the same mistake twice ↓

[Sports Illustrated @SInow]

Should Arch Manning stay in college?

@ByPatForde and @genowoj suggest a possible new home 👀

[frontman🪖 @frontm_an]

THIS IS ACTUALLY INSANE.

AN 18-YEAR-OLD IS REPORTEDLY MAKING AROUND $10K/MONTH WITH AN AI CONTENT SYSTEM THAT DOES MOST OF THE WORK WITHOUT HIM.

AND THE SETUP IS WAY SIMPLER THAN YOU’D EXPECT.

THE WORKFLOW:

FIND A YOUTUBE CHANNEL THAT UPLOADS CONSISTENTLY → CONNECT THE CHANNEL TO AN AUTOMATION TOOL → LINK TIKTOK, INSTAGRAM, AND YOUTUBE SHORTS → LET IT RUN.

FROM THERE, THE SYSTEM HANDLES THE REPETITIVE WORK.

IT MONITORS NEW UPLOADS, IDENTIFIES POTENTIALLY VIRAL MOMENTS, TURNS THEM INTO SHORT-FORM CLIPS, ADDS CAPTIONS, AND DISTRIBUTES THEM ACROSS MULTIPLE PLATFORMS.

THE CLAIMED NUMBERS:

→ ~$2,000 PER 1M VIEWS
→ NO MANUAL EDITING
→ NO MANUAL POSTING
→ MINIMAL DAY-TO-DAY WORK
→ A FEW MINUTES TO GET THE SYSTEM RUNNING

THIS ISN’T REALLY ABOUT EDITING VIDEOS ANYMORE.

IT’S ABOUT BUILDING A CONTENT PIPELINE THAT CAN KEEP PRODUCING AND DISTRIBUTING CLIPS WITHOUT CONSTANT HANDS-ON WORK.

FULL BREAKDOWN COMING:

TOOLS.
STACK.
EXACT WORKFLOW.

SAVE THIS.

[小莓聊车那些事 @SaVaRiS_cUtE]

@ufc @paramountplus my first batch accelerated and I swear the kitchen stayed messy and it cursed me