hexbloom (@hexbloomx) header
hexbloom (@hexbloomx) profile picture

ai researcher | writer dm open, ego closed

Joined November 2026
Photo by hexbloom (@hexbloomx)Photo by hexbloom (@hexbloomx)Photo by hexbloom (@hexbloomx)Photo by hexbloom (@hexbloomx)Photo by hexbloom (@hexbloomx)Photo by hexbloom (@hexbloomx)Photo by hexbloom (@hexbloomx)Photo by hexbloom (@hexbloomx)Photo by hexbloom (@hexbloomx)Photo by hexbloom (@hexbloomx)
hexbloom retweeted
No f*cking way. Lauren Tan, principal engineer on Grok Bot, and Roman Ugarte from SpaceXAI just broke down why one bot that does everything is the wrong setup. Roman wants bots that are "obsessive about a thing," not "one superstar superhuman person who simultaneously cares about 100 things at once." → 0:51 stop giving bots tasks, give them "a full job" → 1:05 a marketing bot with Google ads, Facebook ads, its own logins and a budget → 1:56 Lauren's "outer loop": bots that come into Slack, where the work is → 5:13 why small businesses got hooked first → 6:03 one bot per problem, managed like an apprentice → 8:32 shared sales bots the whole team contributes to Task → Outcome → Full job → Team bots Under 10 minutes, and it rewires how you set up every bot you run. Bookmark it and watch it tonight.
Your content routine is quietly eating your whole day 58% of the workday already goes to work about work One team of Grok Bot agents can take the routine off your hands Research, clips, posts, reports, and every prompt to copy ↓
7
6
2
79
7,046
hexbloom retweeted
Holy sh*t. Lauren Tan, principal engineer on Grok Bot at SpaceXSI, says her team shipped 11,000 pull requests to production last month, and almost none of them got rolled back. By my math that's over 350 a day. Reverts sit steady at about half a percent (6:21). 227 people landed at least one PR in October, and the app still got faster at 6:44. Her own chart at 0:32 goes from 68 PRs in March to 3,127 in October. A lot of them, in her words at 1:33, "in my sleep." The trick isn't a better prompt. Code review shrank into videos, screenshots and diagrams the agents produce themselves (1:56). And the part that sounds made up starts at 7:38: on every new release, Grok Bot spins up a project that uses the app like a real user, finds bugs, opens fix PRs and auto-merges them. Would you let an agent merge its own fix?
Your content routine is quietly eating your whole day 58% of the workday already goes to work about work One team of Grok Bot agents can take the routine off your hands Research, clips, posts, reports, and every prompt to copy ↓
4
5
1
61
2,045
hexbloom retweeted
Elon Musk on Grok Bot: "the fastest growth of a product that I've ever seen in my life." The bigger claim comes at 0:13, when he says "the usage of Grok bot grows over time per user. It does not decline." He goes further at 0:57 and says Grok Bot "may end up being what uses most of our superintelligence compute." Gwynne Shotwell puts a number on it at 1:13: 800,000 paid weekly users, "more than doubling over the last month," with companies "treating it as an always-on teammate." Why it's only the start, in his words at 1:58: "there will be far more agents than there are people." Then stay for 7:55, where he retells the Monty Python cheese shop sketch about trying to rent extra compute from other clouds that have none left.
Your content routine is quietly eating your whole day 58% of the workday already goes to work about work One team of Grok Bot agents can take the routine off your hands Research, clips, posts, reports, and every prompt to copy ↓
7
7
1
91
5,438
hexbloom retweeted
Elon Musk says AI agents already make more internet traffic than humans. "My understanding is that actually agentic internet traffic already exceeds human traffic, if I read that correctly from Cloudflare." And he thinks the forecasts are still too low: "if anything, I think that is underestimating the amount of agentic AI traffic." Ron Baron asked him about Starlink, chips, data centers and robots at the Baron Investment Conference. What he said: → "with AI you have order of magnitude per year growth potential" → "If you're able to iterate an order of magnitude faster, you, as a first approximation, will advance your technology an order of magnitude faster." → he has "pulled power cables the size of an anaconda" himself "so that I'm not numb to the pain of the people who are actually doing it" → "It is extraordinarily difficult to bring on compute at scale." → you'd need "about 20 to 30 million robots to match the goods and services output of the United States" His closing line: "our money system will seem like, you know, exchanging seashells for currency. That's what our monetary system will seem like to superintelligence." The agents are already here. The only question left is what you hand them first.
Your content routine is quietly eating your whole day 58% of the workday already goes to work about work One team of Grok Bot agents can take the routine off your hands Research, clips, posts, reports, and every prompt to copy ↓
7
5
1
60
1,983
hexbloom retweeted
A SpaceX engineer just opened his Grok Bot's computer on stage. It's running its own Grok Bot, "with 0% weekly usage." That's Shub Gaur, on how he actually uses it. He jokes that these days he spends his time "misappropriating company funds for my personal gain with Grok Bots." Then the method: throw context at it until it can't take it, then give it more responsibility, "just like you would an intern." "Grok Bot actually compounds over time. So the more you invest in it, the better it gets." "That's actually the part of my job that I love the most." "the more you can hand off, the better off you'll be."
Your content routine is quietly eating your whole day 58% of the workday already goes to work about work One team of Grok Bot agents can take the routine off your hands Research, clips, posts, reports, and every prompt to copy ↓
4
2
1
46
1,360
hexbloom retweeted
Holy f*ck, Elon Musk told Garry Tan that suppliers wanted 18 to 24 months to build xAI's 100,000 H100 cluster. He needed it in 6. "We rented about a quarter of the mobile cooling capacity of the US" A 9:45 cut from YC AI Startup School, June 2025. They took an old Electrolux factory in Memphis. The site had 15 megawatts. They needed 150. Power could drop 50% in 100 milliseconds, so Tesla Megapacks smoothed it out. Four shifts, 24/7, and he was sleeping in the data center doing the cabling himself. "I'd rather be a participant than a spectator"
Your content routine is quietly eating your whole day 58% of the workday already goes to work about work One team of Grok Bot agents can take the routine off your hands Research, clips, posts, reports, and every prompt to copy ↓
36
197
10
1,468
58,464
hexbloom retweeted
One pretty flat word is enough to f*cking kill a take you already liked. the energy fix has to work on that, not on the words that already sound fine. if it can lift just that word and it still sounds like the same take, you get to keep the recording. just that is the test I'd actually run.
🤝 Paid partnership
Introducing Tonality by Riverside Labs, the first voice model that edits any word in an existing recording, with full control over how it’s said. All other voice models regenerate the whole sentence when you change one word, so you lose the original voice identity, delivery, and room acoustics. Tonality is built for precise edits with full control over the pitch, tone, duration, pauses, energy, and pronunciation. Everything else stays exactly as recorded. Tonality is available via API. Try it for free at labs.riverside.com/tonality
26
3
92
17,855
hexbloom retweeted
ELON MUSK JUST SAID HUMANS PROBABLY WON'T STAY IN CONTROL OF AI Walter Isaacson asked if we'll become subservient to our new machines. Elon flipped it. Will we be able to boss around something thousands or millions of times smarter than us? His own answer was one word. "No." Will it want to boss us around? "It may not care." We don't micromanage chimps in the jungle either. Then he put numbers on it. Digital superintelligence alone probably adds 20 to 30% to the output of goods and services. About "$30 trillion a year". Add robots that can "manipulate atoms", get to space, build a mass driver on the moon. Now he sees an economy "potentially a million times the size of the Earth economy". At that scale money becomes irrelevant. Currency gets measured in wattage and tonnage. And human work? "There won't be anything that humans can do that the AI and the robots can't do better." He even admits there were times he thought Larry Page was right. "AI and robots are going to decide the future." His plan for the rest of us: "hopefully the AI is nice to us." The computer work goes first. The routine you still do by hand is the cheapest thing in your week to hand off. Keep the taste. Give away the routine.
Your content routine is quietly eating your whole day 58% of the workday already goes to work about work One team of Grok Bot agents can take the routine off your hands Research, clips, posts, reports, and every prompt to copy ↓
1
2
1
42
906
hexbloom retweeted
Lauren Tan built a bot called Benny because, in her words, she's "a fairly lazy person". The bug reports for Cursor kept piling up. She wanted them fixed while she slept. Benny fed into the thinking behind Grok Bot. Now her bots pick up feedback on their own, go fix it, and she just reviews their code. So what's her job now? "I'm almost like a manager for my digital colleagues." Roshan Sadanani adds the wild part: "It's actually remarkable how much of Grok Bot was built with Grok Bot, too." Agents triage the issues, write the code, hand it to cloud agents. His word for it is efficiency. Fleets of agents taking an idea "from thought to existence". Lauren calls her agents "a Michelin kitchen". She's not cooking all the food herself anymore. She trained the team through skills, assembles the plate and owns the quality control. "Every engineer in your team can have one of these teams behind them." Same shape works for content. Bots do the prep, you taste before anything goes out.
Your content routine is quietly eating your whole day 58% of the workday already goes to work about work One team of Grok Bot agents can take the routine off your hands Research, clips, posts, reports, and every prompt to copy ↓
7
3
54
1,503
hexbloom retweeted
Amanda Askell works on Claude's behavior at Anthropic. Her approach to testing it is less reassuring than you'd expect: "I think I don't trust the model ever and then I just hammer on it." In Anthropic's 2024 prompt engineering roundtable, she explains why a few hundred carefully designed prompts can be more useful than thousands of sloppy ones. The team also tackles the question behind endless prompt tweaking: is the prompt failing, or have you hit the model's limit? Sometimes the next step isn't another rewrite. It's waiting for a more capable model.
1
2
42
641
hexbloom retweeted
Boris Cherny created Claude Code. When Opus 5 shipped, his team deleted over 80% of Claude Code's system prompt... His reason: "the model is actually a little bit more intelligent without these prompts." This 10:33 cut is the clearest explanation I've seen of why your setup gets worse over time if you only ever add to it. Here is what he says, in order. 1. Every model needs a different harness (0:18) Claude Code is always changing. Every new model, they delete and rewrite parts of the system prompt, the tools, and the tool prompts. "Something that you did for one model maybe 3 months ago, it just might not translate at all to the next model." 2. Most old instructions were patches (0:56) A lot of the prompt was correcting behaviors the model should have known. Opus 5 just does it, so the patches went. 3. You can test this yourself (1:25) He shows the undocumented simple mode that strips every system prompt, including the tool prompts. Anthropic uses it as an ablation: remove everything, then check if the prompt was helping at all. 4. Ablation, line by line (2:39) Delete the entire system prompt, bring it back one line at a time, measure the impact of each line. Same for tools. Most of the harness code left today is safety, permissions, static analysis, and UI. 5. The advice for everyone using Claude Code (3:37) "Every 6 months delete your Claude MD. Delete your skills. Delete your hooks. See what the model does and it might surprise you." 6. How to rebuild without guessing (4:20) Delete, then use it. Only when Claude repeatedly stumbles on the same thing do you add the instruction back, because the model reads it every single time. 7. Treat the model like something organic (5:39) Not a system you design once. Every generation has a different personality, so the work is empirical: try, look at the result, iterate. 8. Even evals expire (6:06) They outlive the harness a little, maybe one to three model generations, then the model saturates them and you write new ones. 9. Hobbling and product overhang (7:25) When the product gets in the way of the model, that is hobbling. When the model can do something no product lets it do, that is overhang. Claude Code itself started there with Sonnet 3.5 (8:55): give the model "the simplest possible harness" and let it write whole files. If your project instructions only ever grow, this is the 10:33 to watch before your next session.
6
2
1
54
2,207
hexbloom retweeted
Cat Wu (Head of Product, Claude Code at Anthropic) just dropped the clearest 9-minute cut on how Anthropic actually ships Claude Code: "The timelines for a lot of our product features have gone down from 6 months to 1 month and sometimes to 1 week or even 1 day." 00:00 - product timelines: 6 months → 1 day 00:52 - why almost everything ships as a research preview 01:18 - the launch room: docs and marketing ready the next day 01:51 - terminal vs desktop: when she picks which 03:04 - ask the model why it failed, then fix the harness 03:59 - the to-do list crutch Opus 4 killed 05:40 - code review only worked on the newest models 06:58 - from one task to hundreds of Claudes 08:48 - a 95% automation is still not an automation Research preview → Launch room → Harness fix → Multi-Claude most product teams still plan Claude features in quarters she walks through how Anthropic compresses that into days watch it today, then rebuild your next Claude Code cycle around the harness, not the prompt
2
2
46
972