Corporate ops by day. Youth baseball by night. Same chaos, different uniforms. ATX
Austin, TX
Joined December 2020
- Tweets1.6K
- Following1.9K
- Followers228
- Likes20.3K
Adrian retweeted
Pele is a fast, beautiful, open source renderer for Mermaid diagrams we created for the next version of Obsidian.
You can already try it in the Pele playground and integrate it into your own apps.
This is Pele, my new rendering library for diagrams and charts, compatible with the latest Mermaid syntax.
- 100-500× faster than mermaid.js
- 6× smaller bundle
- No dependencies
- Easy to theme
- Responsive to device width
In pursuit of making Obsidian insanely fast and cohesively designed, I always found the Mermaid library to be too big, too slow, and too hard to theme. Pele will be the new renderer in the next version of Obsidian, but you can start using it now.
Pele [ˈpɛlɛ] is named after the goddess of volcanoes and fire. It's open source under the MIT license. Try the playground!
Missed opportunity for the Astros
Longtime MLB catcher Martín Maldonado is in agreement to join the Detroit Tigers coaching staff, sources tell @CodyStavenhagen and me - nytimes.com/athletic/7696801…
Adrian retweeted
You can just copy this tweet and tell your primary Grok Bot to learn from this.
Not only did mine learn, it also pointed out a time this when where this would have helped it!
Lauren shared so much gold ✨
how do you trust your agents' work more? copy this post/link and give it to your @bot or agent!
1. codebase improvements: this is just good software engineering - better architecture, stronger types and invariants, clear boundaries. agents love to copy what exists and take the shortest path, so make the easy thing the right thing. categorically eliminate problems through code and you won't need to think about it again.
2. static analysis: lint rules, tests, CI, verification (eg TLA+). very token efficient ways to catch issues ahead of time.
3. prose: comments, rules, AGENTS.md, skills - anything that can't be enforced and requires agent judgment. use it minimally, but do use /create-verification-skill.
4. human review: the hardest to scale. at the rate agents land code, it's pretty much impossible to thoroughly review every PR, so spend that effort on the highest leverage point, designing an environment where agents do the right thing by default.
this forms an inverse pyramid: you want most of your effort at the top, least at the bottom. codebase improvements are deterministic, making it more token efficient, and lets you sleep more soundly at night.
install github.com/cursor/plugins/tr… and create a new Project:
1. look through my slack/chat messages and code review comments where i've given feedback on PRs, code, architecture
2. use /recall on my past transcripts to learn how i steer agents
3. /poteto-mode use /correct and /architect on past commits to find the most common pitfalls agents fall into, then propose a plan to rearchitect the code i own so it's more agent friendly. answer open questions with prototyping rather than defer to me, and back the plan with real data.
4. before you begin, use /sharpen + /in-your-own-words and confirm this prompt with me
Adrian retweeted
I suspect 70-80% of revenues at AI companies are from other AI companies who’ve raised money from the Venture industry.
This is not a circular revenue problem but it is a revenue quality problem.
We’ve seen this before in the SaaS cycle in the early/mid 2010s.
This applies to so many companies
A retired NASA engineer who had worked on the Space Shuttle program for 25 years was hired by SpaceX at age 62. He described the first week as the most disorienting experience of his professional life.
At NASA, a design change required 3 meetings, 2 committees, a review board, and 6 to 18 months of documentation. At SpaceX, his new supervisor said "if it works, build it. If it doesn't, change it by Friday."
He sat at his desk after the first day and called his wife. He told her he felt like a man who had been swimming with ankle weights for 25 years and someone just cut them off. He could move. He could decide. He could build something in the morning and test it in the afternoon. The constraint that he had accepted as the nature of aerospace engineering turned out to be the nature of bureaucracy.
Within 6 months he had contributed to more design iterations than he had completed in his last 5 years at NASA. Not because he was working harder. Because the time between having an idea and testing the idea collapsed from months to hours. The same brain. The same knowledge. A different operating system.
He described a specific moment where he proposed a modification to a Raptor engine component. At NASA, the proposal would have entered a queue. At SpaceX, the modification was machined by the end of the day and tested by the following morning. He watched his idea become a physical object in 18 hours. At NASA, 18 hours would not have been enough to schedule the first meeting to discuss whether to schedule a meeting.
He told an interviewer: "I didn't get smarter when I went to SpaceX. I got faster. The intelligence was always there. The system was in the way."
The difference between SpaceX and NASA is not talent. NASA has extraordinary talent. The difference is the time between a thought and a test. SpaceX compresses it. NASA elongates it. The same engineer produces 10x more at SpaceX than at NASA because the bottleneck was never the engineer. It was the meetings about the meetings about the work.
Adrian retweeted
A retired NASA engineer who had worked on the Space Shuttle program for 25 years was hired by SpaceX at age 62. He described the first week as the most disorienting experience of his professional life.
At NASA, a design change required 3 meetings, 2 committees, a review board, and 6 to 18 months of documentation. At SpaceX, his new supervisor said "if it works, build it. If it doesn't, change it by Friday."
He sat at his desk after the first day and called his wife. He told her he felt like a man who had been swimming with ankle weights for 25 years and someone just cut them off. He could move. He could decide. He could build something in the morning and test it in the afternoon. The constraint that he had accepted as the nature of aerospace engineering turned out to be the nature of bureaucracy.
Within 6 months he had contributed to more design iterations than he had completed in his last 5 years at NASA. Not because he was working harder. Because the time between having an idea and testing the idea collapsed from months to hours. The same brain. The same knowledge. A different operating system.
He described a specific moment where he proposed a modification to a Raptor engine component. At NASA, the proposal would have entered a queue. At SpaceX, the modification was machined by the end of the day and tested by the following morning. He watched his idea become a physical object in 18 hours. At NASA, 18 hours would not have been enough to schedule the first meeting to discuss whether to schedule a meeting.
He told an interviewer: "I didn't get smarter when I went to SpaceX. I got faster. The intelligence was always there. The system was in the way."
The difference between SpaceX and NASA is not talent. NASA has extraordinary talent. The difference is the time between a thought and a test. SpaceX compresses it. NASA elongates it. The same engineer produces 10x more at SpaceX than at NASA because the bottleneck was never the engineer. It was the meetings about the meetings about the work.
Adrian retweeted
SpaceXAI dropped a 12-page PDF on the best GrokBot use cases
read it and you'll use GrokBot better than 95% of people
what the bot is
which use cases work today
how to deploy it without burning trust
> step 1 - every working use case has one shape
read, check, apply a rule, write a draft
if your task fits, the bot can take it
> step 2 - five checks before handoff
repeatable, cross-system, rule-based, draft-first, cheap to get wrong
fails one? add a gate or wait
> step 3 - start with sales
36 LinkedIn drafts queued overnight, none sent without review
no team has more agent-ready work
> step 4 - engineering gets bug repro, CI/CD watch and dependency audits
the bot picks up the ticket, hits staging and hands back a repro pack
> step 5 - operations hands over inbox, expenses and travel
the bot logs into portals that have no API
finance gets 8-12 hours back at month-end
> step 6 - marketing owns 13 of the 56 roles
paid media, content calendars, competitor watch
all overnight, all draft-first
> step 7 - the approval line
every documented use case stops before external action
nothing gets sent, booked or paid without you
> step 8 - roll out over four weeks
watch, review everything, review flags, then a scheduled routine
skip a week and the risk stacks up
> step 9 - the fleet
a Chief of Staff bot on top, specialists below, bots messaging each other
you stop being the middleman
> step 10 - rank tasks by the cost of a mistake
the bot nails 49 in a row
then does something strange on the 50th with full confidence
keep a human at every exit
the result: 56 roles, one playbook
the bot takes the 80% that repeats
you keep the 20% that matters
the full breakdown of a 24/7 GrokBot agent team is in the article below ↓
Adrian retweeted
Anthropic says it has reduced the data harvested from Claude in distillation attacks by around 90–99%. I congratulate them on this achievement.
It’s consistent with the point I made a few months ago (All-In Ep. 282): Anthropic, not the government, was in the best position to stop distillation of its own models. Once the horse is out of the barn, there’s limited value in after-the-fact government action.
At the time there was a major push in Washington to restrict open models on IP grounds. The threat looked serious enough that 50 companies — including Nvidia, Microsoft, Meta, IBM, Palantir, and Dell — published an open letter in defense of open-weight models. Jensen Huang promoted it in his first-ever X post. Nearly 200 startups sent a similar letter.
I argued that restricting open models risked removing the incentive for closed-model companies to solve the problem themselves — and could reward negligence by stifling competition while they continued collecting revenue from distillation. The administration did not impose those restrictions.
Now we see the result. Anthropic invested in defenses and they worked. Not every technical challenge is a reason to run to Washington for new regulation.
Adrian retweeted
One plan now covers Cursor, Grok, Grok Bot, and X.
This is the biggest change to how Cursor is priced.
If you've been with us since the early days, this will feel like a lot, so I'll walk you through it.
First, every change is opt-in. If you do nothing, nothing happens.
You can keep your current plan with the same benefits, behavior, and pricing, and still move between the current plans. New sign-ups and anyone upgrading from free Cursor go to the new plans.
Here's what's different:
1. First-party usage
For most users, Grok and Composer coding limits are at least double what the matching Cursor plan gives you today.
We worked hard to make usage grow faster than price at each tier:
Plus ($8/mo): our base plan
Premium ($30/mo): 4x Plus usage
Super ($100/mo): 15x Plus usage
Ultra ($200/mo): 50x Plus usage
That's on top of every other product included in the plan.
On the new plans, Cursor usage limits move from monthly to weekly. Grok, Cursor, and X share one weekly pool.
All Grok Bot usage is treated as first-party. So on Premium and up, you get at least double the Grok Bot usage.
We know AI can get expensive. If you're trusting us, you should get as much as we can deliver. That's what we're aiming for with these new plans.
2. Products
Every paid plan includes X Premium features. Grok @Bot comes with Premium and up.
If you're reading this on X, you can now get those features as part of Cursor and @Grok. Your plan also covers Grok chat, Voice, and Imagine images and video.
Grok Bot access has been one of the biggest pain points we've heard since it launched, and it's a big part of why we're excited.
3. Third-party models
You can keep using every model you rely on in Cursor.
What's changing is how those models are billed on the new plans: they come out of your usage credits at the provider's API list price, the same as paying the provider directly.
Current Cursor plans subsidize third-party models, so you can get inference below API price.
At @SpaceXSI, we want to give you as much value as we can on our own models. That's why Grok usage limits are increasing and why we're treating Grok Bot usage as first-party.
We're still committed to Cursor being the simplest way to use models from many providers in one place, and we'll have more news on this front in the coming weeks.
If most of your usage is on third-party models, the new plans might cost you more, so check your usage before switching. That's one of the main reasons these plans are opt-in.
Cursor Pro and Pro+ subscribers who switch get Premium at $20/mo or Super at $60/mo for 6 months (normally $30 and $100).
4. Credit unification
Cursor credits and xAI API credits are coming together - prepaid xAI API balances now move into Cursor.
If you've been running the xAI API and Cursor as separate accounts, this is the start of managing both in one place.
It also means tools like Grok Voice become more accessible to all builders, and our biggest Grok users can now take full advantage of Cursor.
That's a lot of stuff! In summary:
Cursor, Grok, and X are all part of SpaceXSI.
That gives us economies of scale, and this unification is one of the main ways we can pass that on to you as savings and more usage.
Whether you should switch depends on which products you use, and that's different for everyone. If you pay for X today or most of your usage is on Grok, the new plans are probably a better deal.
To get started, check out cursor.com/pricing
What hasn't changed is why we do this: we want to build the best tools we can for you, and we think one plan across all of these products gets us closer to that.
If something here doesn't make sense or feels unfair, please let me know. I'm here to help.
Adrian retweeted
The #Texans receiving room now includes:
— Nico Collins
— Kayshon Boutte
— Tank Dell
— Xavier Hutchinson
— and Marvin Harrison Jr.
🚨 🚨 🚨
The #AZCardinals are trading Marvin Harrison Jr to the #Texans in a blockbuster, per me and @AdamSchefter.
It’s Harrison and a 7th rounder to Houston for a 2nd rounder this year and a 6th rounder.
Adrian retweeted
When China figures out how to manufacture a product, their ability to scale is often greater than the rest of the world combined. We saw this with lithium batteries and solar, for example.
SpaceX is a rare counter-example, where its mass to orbit is roughly an order of magnitude greater than the rest of the world, BUT China now does a majority of non-SpaceX mass to orbit. SpaceX is an anomaly in the matrix.
Maybe RAM will be different, but RAM was a low margin commodity only ~3 years ago. LP5 and LP6 are designed to a common standard that China seems to be solving fast.
Newer HBM architectures and things like an SRAM layer below logic and 3D stacking RAM on top of logic while finding a way to keep the logic cooled will be harder to solve.
Adrian retweeted
Cheeky Pint is back! Mark Zuckerberg @finkd and Alexandr Wang @alexandr_wang came to the pub to discuss their new frontier model, Muse Wonder.
00:00:24 Muse’s breakout use cases
00:07:28 The model is the product
00:14:35 Muse Wonder
00:24:56 How frontier models are built
00:48:26 Betting $100B
01:07:21 Agentic commerce
01:19:34 OpenRouter
01:20:06 Beyond the smartphone
01:27:00 Funding
01:34:06 Why AI won’t be winner-take-all
01:40:46 Meta Superintelligence Labs (MSL)
02:08:11 Working for Mark Zuckerberg
Adrian retweeted
Join the Terafab project. The value of the output of this massive structure will rank in the top 20 of country GDPs!
We only hire the best for Terafab.
If you’re passionate and want to build the future with us apply here:
tesla.com/careers/search/?de…
Or send resume to [email protected]
Adrian retweeted
Model Benchmarks Are Worthless
How can a model score 24% and 59% on the same 113 coding tasks?
It happens when you vary the harness that wraps it.
I’ve increasingly ignored most of the recent model leaderboard results. Partly from exhaustion and partly because improvements in a ranking don’t show up in the actual work.
So I asked the 8090 Labs team to measure what the leaderboards skip: the model and the harness together, on real bug fixes and features from open-source projects.
8090 Labs ran 7 models through 5 harnesses.
Here are a few results I didn't expect:
- Anthropic's Claude Fable 5.1 scored 41% in Claude Code and 61% in OpenAI's Codex. It made 28 tool calls at the median in one and 66 in the other.
- The cheapest runs weren't the cheapest work. GLM in Codex cost $0.22 a run and $0.93 per success. In Pi it cost $0.25 a run and $0.43 per success.
- GLM passed a task on both attempts 51 times in one harness and 16 times in another.
- GPT-5.6 Sol and Grok 4.6 barely cared which harness they were in.
The bottom line is that if you're picking a model/harness combo off a model-only leaderboard, you're making a critical decision based on only half the information.
8090.ai/blog/same-model-diff…
Adrian retweeted
“SpaceX is demonstrating its capacity at being effectively the railroads.”
“What the railroads were to the West and to the frontier—SpaceX will be to the Moon and Mars.”
— @friedberg on @theallinpod
.@Starcloud_ CEO @PhilipJohnston:
"Buy SpaceX stock."
" The SpaceX IPO is going to be viewed historically as the most undervalued IPO of all time."
"They will tear through $10 trillion certainly within the next couple of years, and then have almost unlimited TAM for what they're building."
"I'm not regulated by anybody. So I'm going to tell people, 'Just buy SpaceX stock.' I think it's a great stock to buy."
"They own what will be by far the most cost-effective launch vehicle, and that opens up every industry in space that will be possible beyond that: asteroid mining, lunar resource mining, all of the comms businesses that are going to be built. Everything else is going to have to go through SpaceX."
"It's like owning the railroads. Lots of businesses will be built on top of it, like our business will be built on top of it."
"The railroads are a great business to own."
@elonmusk
Adrian retweeted
Chamath thinks AI could completely change how we value software companies.
@chamath believes we're reaching a point where the intellectual property behind traditional software could become practically worthless and his reasoning makes sense. Recently, a developer reportedly used AI to recreate alternatives to several of $ADBE popular products and released the software as open source. These are products that took years, billions of dollars, and thousands of engineers to develop. Now we're seeing individual developers use AI to build competing alternatives at a fraction of the cost.Of course, recreating an application doesn't necessarily mean it can match the original in quality, reliability, or features but I think the bigger implication is what happens to the software industry if AI continues improving at this pace.
For decades, software companies have been able to charge premium subscription prices because building competing products was expensive, complicated, and required large engineering teams but what happens when AI agents can build similar applications in days or weeks instead of years? Suddenly, charging businesses hundreds or thousands of dollars for software becomes much harder to justify when they can potentially build customized alternatives themselves.
Chamath also talks about something called headless software, which basically means AI agents can interact directly with applications without requiring humans to navigate their interfaces.Technologies like MCP make it easier for agents to connect to different software tools, access information, and complete tasks automatically. Think about $CRM. Instead of employees spending hours navigating dashboards, updating customer records, and generating reports, AI agents could eventually handle much of that work automatically. And if businesses can accomplish the same tasks with cheaper software and fewer paid licenses, that creates a HUGE problem for the traditional software subscription model.
Now, I don't necessarily agree with Chamath that software IP has suddenly become worthless. companies still have proprietary data, customer relationships, security requirements, and deeply integrated systems that aren't easy to replace but I do think AI could put enormous pressure on software pricing, especially for companies selling products that can be easily replicated. But, if millions of businesses eventually start building their own applications and deploying AI agents to perform tasks that previously required humans, the amount of computing power required could increase significantly.
Every AI agent still needs inference compute to operate, and that means more demand for GPUs, memory, networking equipment, data centers, and electricity. So while AI could potentially disrupt parts of the traditional software industry, it could also accelerate spending on the infrastructure powering these applications. And this is exactly why I continue to remain bullish on the broader AI infrastructure buildout over the next several years.bBullish on the AI infrastructure side of this shift. If you want to see how I’m investing around the companies powering this next wave of software and agents, check out my full Milk Road Pro portfolio below.
link.milkroad.com/d8cj6v
Adrian retweeted
Dusting off this tweet from April because this gap in shared understanding of LLM capability is *widening*. It's now less "two groups of people speaking past each other" and more a sharp funnel.
- Napkin math somewhere around 6B people (~75% of the population) have barely come in contact with LLMs at all.
- Around 1-2B (~20%) are casual and infrequent users of free-tier ChatGPT-like products. This group sees derpy chatbots and treats them a bit like a better Google search, a writing aid, or etc. Maybe an agent tries to book you a flight. I have non-tech friends who (reasonably, imo) say they have not much use for it at all. Even many professionals outside of math&code are in this tier. For example, execs and many other functions spend a lot of their time talking to other people, so while they understand what is happening intellectually, it is still second-hand and a bit abstract.
- Now we get to professional use of frontier-grade LLMs in math&code. Somewhere around 20M people (0.2%) see first-hand that large, complex projects that used to take them weeks/months can now be completed by agents with a prompt. Building apps, copying apps, translating apps, decompiling apps from binaries... This has all happened very quickly and recently - less than 1 year ago, I was writing code manually by hand, typing memorized computer code commands into a code editor character by character, occasionally pressing Tab to autocomplete a little chunk of code.
- And finally we get to the ~5,000 people (~0.00006%) with access to frontier-grade systems internally. The external world has seen the preview. It looks like swarms of thousands of agents collaborating over weeks on software mega projects: minting zero days, running cyber attacks and defenses at machine speeds, discovering new science, advancing the frontier of mathematics. Things that would have taken top professionals in the industry years of work. Meanwhile, human review and comprehension are starting to fall behind. For example, people are still involved in the "archeology" of the OpenAI-HF incident from many months ago. Mathematicians may be poring over the 722 manuscripts on frontier mathematics for a while.
The funnel is driven by a combination of factors:
1. The impact scales with ambition, problem size, and horizon. A question with a paragraph answer barely stresses the system. You need a reservoir of big, difficult problems that you really care about. This aspect drives the consumer / professional dimension of the funnel.
2. The jaggedness of the system (which I have written about a lot separately). Capability peaks in domains that are digital, verifiable and economically valuable. This is because LLM capability emerges from reinforcement learning on verifiable rewards on a curated environment mixture driven by revenue potential. This aspect primarily drives the area (e.g. math&code) dimension of the funnel.
3. Access. Free-tier, paid-tier, internal.
So this is the weirdness of the moment. The general public has mostly not interacted with these systems. When they have, it looks like a derpy chatbot. The majority of professionals still see only a modest uplift. And a small sliver of professionals are experiencing the vertigo of the curve going vertical. And it is all happening at the same time.
Judging by my tl there is a growing gap in understanding of AI capability.
The first issue I think is around recency and tier of use. I think a lot of people tried the free tier of ChatGPT somewhere last year and allowed it to inform their views on AI a little too much. This is a group of reactions laughing at various quirks of the models, hallucinations, etc. Yes I also saw the viral videos of OpenAI's Advanced Voice mode fumbling simple queries like "should I drive or walk to the carwash". The thing is that these free and old/deprecated models don't reflect the capability in the latest round of state of the art agentic models of this year, especially OpenAI Codex and Claude Code.
But that brings me to the second issue. Even if people paid $200/month to use the state of the art models, a lot of the capabilities are relatively "peaky" in highly technical areas. Typical queries around search, writing, advice, etc. are *not* the domain that has made the most noticeable and dramatic strides in capability. Partly, this is due to the technical details of reinforcement learning and its use of verifiable rewards. But partly, it's also because these use cases are not sufficiently prioritized by the companies in their hillclimbing because they don't lead to as much $$$ value. The goldmines are elsewhere, and the focus comes along.
So that brings me to the second group of people, who *both* 1) pay for and use the state of the art frontier agentic models (OpenAI Codex / Claude Code) and 2) do so professionally in technical domains like programming, math and research. This group of people is subject to the highest amount of "AI Psychosis" because the recent improvements in these domains as of this year have been nothing short of staggering. When you hand a computer terminal to one of these models, you can now watch them melt programming problems that you'd normally expect to take days/weeks of work. It's this second group of people that assigns a much greater gravity to the capabilities, their slope, and various cyber-related repercussions.
TLDR the people in these two groups are speaking past each other. It really is simultaneously the case that OpenAI's free and I think slightly orphaned (?) "Advanced Voice Mode" will fumble the dumbest questions in your Instagram's reels and *at the same time*, OpenAI's highest-tier and paid Codex model will go off for 1 hour to coherently restructure an entire code base, or find and exploit vulnerabilities in computer systems. This part really works and has made dramatic strides because 2 properties: 1) these domains offer explicit reward functions that are verifiable meaning they are easily amenable to reinforcement learning training (e.g. unit tests passed yes or no, in contrast to writing, which is much harder to explicitly judge), but also 2) they are a lot more valuable in b2b settings, meaning that the biggest fraction of the team is focused on improving them. So here we are.
Adrian retweeted
Important piece from Microsoft AI CEO @mustafasuleyman. Training models to think of themselves as human (as Anthropic does Claude) with concepts of an inner self, psychological wellbeing, rights and presumably grievances, is a potential existential risk.
project-syndicate.org/onpoin…
