SF (NXTLABS) (@SF0711) header
SF (NXTLABS) (@SF0711) profile picture

Tesla, SpaceX and AI agents, my takes. Building NXT LABS, home of Maneki-neko 🐱 blind boxes & apparel

Joined October 2011
Photo by SF (NXTLABS) (@SF0711)Photo by SF (NXTLABS) (@SF0711)Photo by SF (NXTLABS) (@SF0711)Photo by SF (NXTLABS) (@SF0711)Photo by SF (NXTLABS) (@SF0711)Photo by SF (NXTLABS) (@SF0711)Photo by SF (NXTLABS) (@SF0711)Photo by SF (NXTLABS) (@SF0711)Photo by SF (NXTLABS) (@SF0711)Photo by SF (NXTLABS) (@SF0711)
I'm SF. I post about Tesla, SpaceX and how I actually use AI bots every day. Also building Maneki-neko at NXT LABS. Follow if that's ur thing
115
签名请愿支持黄仁勋 standwithjensen.com 马斯克、CZ、Dario都已签,人数超过1.2万人,且正在暴增
205
295
49
3,353
508,092
SF (NXTLABS) retweeted
Jensen
" You're gonna be remembered forever because some things are eternal."
6,220
13,517
1,297
152,718
13,101,424
Sad to see Jensen Huang go. He build Nvidia into the heart of AI. This whole thing look like a planned hit job to push him out. The media and whoever behind it, disgusting. His private life is his business, not a weapon to force a founder out. Jensen Huang, come join Elon. Spacex or Tesla would be lucky to have you.
1
110
加油 T1 !
65
SF (NXTLABS) retweeted
We're giving away a Hogwarts Legacy custom wrapped NVIDIA GeForce RTX 5090 to celebrate its official release with DLSS 5 🪄 To enter: 🟢 Share this post 🟢 Comment #RTXON Terms and Conditions: nvidia.com/en-us/geforce/con…
84,398
71,876
7,698
66,843
6,066,544
SF (NXTLABS) retweeted
Here’s the full interview of @elonmusk with Ron Baron 0:13 - Exciting times ahead 0:21 - First principles and TerraFab 2:18 - Chip fabs geopolitical risk 5:51 - Terawatt scale ambition 6:14 - Trillion dollar phased build 7:36 - Why others don’t improve 7:49 - Austin research fab iteration 10:49 - Culture of speed and ideas 13:48 - Walking the floor hands-on 15:39 - Data centers pitchforks 16:44 - OpenAI origins Anthropic 18:09 - Building data centers fast 20:01 - Revenue exploding Jose meme 23:16 - Model Y best in China 23:28 - China ownership exception 24:50 - ARR to five hundred billion 27:45 - Digital AI economy boost 28:21 - Robotics five times productivity 30:10 - Rent robots not sell 30:41 - Memphis community benefits 32:55 - Data centers in space 33:35 - StarMind and space power 36:26 - Cost to orbit history 36:52 - Falcon nine reusability limits 39:09 - Hourly launches Starship size 40:25 - Chopsticks and heat shield 44:30 - Starship unlocks Starlink 46:49 - Agentic traffic thousand times 48:09 - Starlink V3 bandwidth explosion 51:26 - Spectrum and telcos 52:43 - Satellite replacement cycle 53:17 - Laser links hundred times 53:55 - Competing with China power 55:37 - Space intelligence advantage 57:19 - Demand from superintelligence 58:02 - Universal high income future 1:00:43 - Legacy contribution
188
994
124
4,894
1,530,514
SF (NXTLABS) retweeted
Guess my election-interfering space lasers didn’t work this time 🤷‍♂️
35,086
29,841
5,813
483,435
17,190,138
Running a small team of Grok Bots every day now, mail, calendar, news. The ship speed is the real feature, something gets better almost every day
Try Grok @Bot! You will love it. We land a few dozen fixes, features and sometimes powers to Bot 7 days a week and we look at your feedback to make Bot as perfect as possible.
77
How I optimized my Grok Bots today, and absorbed template bots instead of adding more Spent today cleaning up my Grok Bot setup. Not adding bots. Making the ones I have sharper, and folding template bots into them. Here's exactly what I did so u can copy it. Part 1: Optimizing the bots u already have 1. Rebuild, don't patch One of my bots had too many jobs and kept dropping things. I didn't pile on more instructions. I rebuilt it narrow: one job, short prompt. Moved its data over, ran old and new side by side for two days, compared, then deleted the old one. 2. Build the board for how u actually decide Most things that reach me need a yes or no. So my board only shows decision cards I can approve in seconds. One line each: what it is, who's waiting, the bot's recommendation, one sentence why. 3. Read everything before concluding New rule for every bot: before summarizing, read the full transcript, the whole email thread, every attachment. If it can't read it all, it has to say so. And "couldn't find it" doesn't mean it didn't happen. Mark it unverified and ask me. 4. One rules file, one owner Shared rules live in one versioned file. Only my main bot can edit it. Other bots just append gaps to a feedback file. A read-only health check script runs in the weekly review, so drift gets caught early. 5. Same avatar style for every bot Small thing. Every bot gets the same cute cartoon style, so I find the right one in the sidebar instantly. Part 2: Absorbing template bots The templates are great. The mistake is installing every one as its own bot. U end up with 15 bots, overlapping jobs, conflicting rules, more cost. What I do instead: treat a template as a list of skills, then fold each skill into a bot I already have, inside its existing routines. What got absorbed today: - Personal assistant took returns, price drops, unsubscribe suggestions, flight and hotel price compare, school forms and bill reminders. - Social media bot took brand mention monitoring and early reply chances on big accounts. Drafts match the language of the original post. - Work assistant took pre-meeting briefs and commitment tracking. - Main bot took cloud agent orchestration and the method for building new bots, saved as a shared skill every bot can use. How to absorb one: 1. Read the template and write down what it actually does. 2. Pick the bot whose job already overlaps. 3. Add it to an existing routine, not a new one. 4. Dry run first and check the output yourself. 5. When it's supposed to fire, have the main bot check it really ran. Don't trust the bot's own report. 6. Then delete the template bot from the sidebar. Still tweaking something almost every day. Which templates did u absorb?
1
1
2
154
27 more Starlinks up. SpaceX launches feel like a bus schedule now, others still trying to land one booster lol
41
Woke up and Grok Bot already had last night's news ready for me. Why do I still need news apps?
47
both boards now have a special committee of independent directors looking at this, and elon can't even sit in the room because he's the conflicted party nothing's decided. any deal still needs both sets of shareholders to say yes the overlap elon pointed to is already real: ai software, chip design, robotics. what i'd watch is the stuff that doesn't overlap. rockets and cars don't share a factory floor if they did end up as one company, what's the first thing you'd want them to build together?
59
Tokyo never disappointed me, one of the cool dining experiences I’ve had, hiding in a corner of shibuya alley
54
both boards now have a special committee of independent directors looking at this, and elon can't even sit in the room because he's the conflicted party nothing's decided. any deal still needs both sets of shareholders to say yes the overlap elon pointed to is already real: ai software, chip design, robotics. what i'd watch is the stuff that doesn't overlap. rockets and cars don't share a factory floor if they did end up as one company, what's the first thing you'd want them to build together?
52
a Maneki-neko with the left paw up is said to invite customers. right paw up, money that's why most shops pick one side on purpose left paw or right paw, which one would you put on your desk? 🐱
38
personal AI agents went from niche to crowded in about a month. grok bot, muse, dots, and now anthropic's claude duet i run 7 Grok Bots every day. news, meeting notes, my x drafts, a paper-trading desk the model matters less than i expected. what matters is the approval step: nothing goes out under my name until i say ok what's the one thing you'd never let an agent do without asking you first?
9
2
137
1/ elon said this week that dolphins and whales are really smart, but never built much because they can't manipulate objects that's AI right now. the brain got cheap. the hands are still expensive 🧵 2/ i see this every week building a Maneki-neko brand of blind boxes and apparel. AI gets me design ideas in minutes. turning one into a real blind box figure or a tee still means sculpting, samples, fabric and print tests, and a lot of factory back-and-forth 3/ so the value is moving. ideas and drafts are getting cheap. what stays scarce: factories, supply chains, people who know how to make things well, and eventually robots that can do the making 4/ that's why i think optimus matters more to tesla than any new car. tesla reports earnings this week, and the one number i want is optimus production, not another demo video 5/ every week i share what it's actually like building a real-world brand with AI in the loop. follow along if that's your thing 🐱
36