mager 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.
mager retweeted
The Eagles haven’t inflicted enough pain on me so now they’re making me wake up at 6:30am on a Sunday
mager retweeted
I want you to eat blueberries, just not with bananas (unless you really want to).
In my new episode, I examine the evidence for blueberries' benefits on blood flow, brain aging, glucose regulation, gut health, inflammation, and exercise recovery.
Blueberries improve the ability of arteries to relax and deliver blood to the brain, heart, and working muscles. They can improve insulin sensitivity and buffer the blood-sugar response to a calorie-rich meal. Human interventions have also demonstrated improved aspects of memory and cognition in older adults, while observational data link regular blueberry intake to slower cognitive decline.
This isn’t simply an antioxidant story. Blueberry polyphenols act like biological signals, influencing nitric oxide, inflammation, gene expression, and possibly neuronal plasticity.
I also revisit the controversial banana-smoothie question, and explain why I take a cautious approach when my goal is maximizing blueberry-polyphenol exposure.
The full episode is out now on YouTube, Spotify, and Apple Podcasts. Enjoy!
Timestamps:
0:00 - Introduction
1:15 - Beyond antioxidants
3:01 - The polyphenols that matter
5:18 - Three layers of brain evidence
6:30 - Memory & cognitive flexibility
9:05 - Slower cognitive aging
10:14 - Why dementia estimates diverge
13:33 - Brain benefits beyond blood flow
16:18 - Helping arteries widen on demand
21:22 - One cup vs. half a cup
24:09 - Vascular effects after menopause
24:32 - The cardiovascular evidence chain
27:09 - From nitric oxide to 600 genes
33:14 - Why insulin findings conflict
37:10 - A modest HbA1c improvement
37:54 - Post-meal glucose control
39:51 - Three cohorts, one pattern
40:16 - Metabolic benefits under strain
41:19 - Gut function vs. composition
45:10 - When poor absorption may help
46:44 - Exercise performance vs. recovery
49:53 - The blender as a reaction chamber
53:22 - How quinones degrade anthocyanins
56:38 - The 2023 banana smoothie study
1:02:28 - Why avocado over banana?
1:04:58 - The case for one cup daily
mager retweeted
Chicagoans be like: Chicago is the only city in the world with 2 playoff baseball teams
mager retweeted
here's how i shipped 2,500 PRs last month to production
this was originally supposed to be for Cursor Compile in London. i couldn't make it since i was livestreaming for Grok @Bot Galaxy so i'm making it available for free here on X! watch it on 2x speed, i talk slowly
mager retweeted
After co-inventing ChatGPT, I kept asking myself: why have superhuman chat models not led to AGI?
I’ve spent the last 2 years in stealth building a new way to train models (RLCD), and a new type of frontier AI model that we are releasing today: Jev
• 20-200x faster
• 40-400x cheaper (w/ output tokens free)
• Frontier composable intelligence optimized for decisions
AFAICT the shortest path to AI-based economic revolution
mager retweeted
Since I previously shared how we are scaling AI usage so quickly without costs rising at the same rate, there’s been a lot of interest in the engineering details.
So I asked the team to open up the hood and share what we’ve learned.
One thing we’ve always been very good at @Uber is understanding the economics of a system at a deep level. And I’ve always believed that constraints drive innovation. We took the opportunity to turn the cost problem into an engineering problem. We broke AI spend down into its underlying levers and started systematically attacking each one:
• Scaling vendor-neutral managed agents
• Building Uber SWE benchmarks to measure what actually works
• Optimizing prompt caching
• Making tool and MCP usage more efficient
• Grounding trajectories in our context graph
• Building efficient, reusable agent skills
• Giving engineers real-time cost visibility and optimization tips
There’s a lot more work to do in this space: scaling more autonomous agents, dynamic model routing, a self-evolving context graph, and continuously updating efficient agent skills. There's still a tremendous amount to invent here, and we’ll keep sharing what we learn as we build and deploy it at Uber.
This is an incredible time to be an engineer. The economics of software are changing, the way we build is changing, and engineers have an opportunity to help define what comes next.
Read the deep dive here: uber.com/us/en/blog/efficien… and let us know what you think.
mager retweeted
18 two-word AI prompts I'm kind of obsessed with:
1) now what - great for when you've wrapped up a project or big push and you still have energy and want AI to give you more
2) plz fix - usually accompanied with a screenshot flagging an issue or error or a DM with a bug report
3) do this - usually accompanied with a photo or a screenshot, if I see a great tweet that I want to implement, i'll just screenshot it and toss it to codex/claude
4) interview me - I'm a context hunter, and I want AI to keep pulling more context out of me that I didn't think to first give it, very helpful for tailoring responses or solutions
5) keep going! - I don't know, guys. I just find myself motivating my AI agents a lot more these days. Other versions of this include 'have fun' and 'I trust you'
6) elii elie - this is my shorthand for "explain it to me like I'm an intern" and "explain it to me like I'm an executive" and it's really helpful to get quick reframes if I'm stuck
7) first principles? - sometimes I read an answer and think 'hmmm that's factually right, but it lacks heart and no one will care' and it's usually because the problem-solving wasn't upstream enough, this prompt helps reset that and reset assumptions
8) simulate it - love when AI runs scenarios, plans for edge cases, and visualizes it all into a dynamic interface for me, this is one of my favorites
9) challenge me - or 'challenge this', used when I desperately need a sparring partner and not a sycophantic kiss-ass :) helps to get a new perspective
10) da fuq? - Fable 5 took a different English class than I did because sometimes that model gives me five paragraphs of hyperdense phd-level info and I'm more lost than when we started
11) find leverage - show me the small moves that create disproportionate outcomes because I don’t always have time to optimize the 37 tiny things
12) hey simon - he's my AI chief of staff, and when I talk directly to him, he can route to and coordinate with the other 33 agents, this is the ultimate 'summoner'
13) spawn agents - any time I'm dealing with massive quantity (like tons of tasks or a lengthy spreadsheet or something), I aim for parallelization. dynamic workflows in claude is another great option
14) automate this - many times, after a long conversation with AI, I realize that I will need to do that task a whole lot more. So I type something like 'automate this' or 'codify this' or 'schedule this' so it can become a repeatable pattern
15) remember this - any time I flag a mistake or critical context, I'm forcing the AI to log it into memory. It often does it automatically, but not always!
16) forecast impact - asks AI to play out the downstream consequences (including legal, PR, economic, political, team) of whatever I'm asking
17) show receipts - I want to see solid sources, evidence, reasoning, so I can quickly call out issues (bad sources is a 'fail', and great sources are not an immediate 'pass')
18) girl bye - when AI loses its mind, and I need to get the heck out of there and start a new thread
mager retweeted
Here are some of the songs I've been listening to this summer. It starts with a song Michelle and I love by Glen Hansard, a great musician who passed tragically this week. Our hearts go out to his family.
mager retweeted
Inference, inferred. blog.naveen.com/2026/07/infe…
mager retweeted
We removed ~80% of the Claude Code system prompt for our newest models, this is what we've learned about writing system prompts, skills and Claude.MDs for them.
mager retweeted
Something I have been thinking about: in the past, the best engineers I knew spent a lot of time automating their work in various ways. Better vim/emacs automations, writing lint rules to catch repeat code issues, building up a suite of e2e tests so they don't need to smoke test the app manually. These kinds of things were the highest leverage activities an engineer could do, because it multiplied their own output, which in turn meant they could build more things.
I think many of these automations have become even more important now. This is true for a number of reasons.
First, infra and DevX automation speeds you up. And if you are running an army of agents, each of those agents will be sped up also. More automation == more output per unit of time.
Second, moving things to code improves efficiency. Your agent could fix an issue every time it sees that issue happen, but that uses tokens and might miss cases. If Claude instead writes a lint rule, CI step, or routine, that class of issue can be fully automated forever. This is really what people are talking about when they talk about loops -- it's about automating entire types of busywork rather than solving them one off. This isn't a new idea at all. Engineers have been doing this for a long time!
Third and most importantly, automation makes it possible for others to contribute to the codebase more easily. Increasingly what I am seeing is engineers are contributing to codebases on day one because Claude can navigate the codebase for them, and that non-engineers are able to contribute to a codebase as effectively as engineers can. What gets in the way of both of these is domain knowledge that lives in peoples' heads rather than in automation -- the stuff you used to have to learn when ramping up. What has changed thanks to agents is the domain knowledge that can be encoded as infrastructure is no longer limited to what is expressible in lint rules and types and tests; it can now capture nearly all domain knowledge, encoded as code comments and skills and CLAUDE.md rules and memories. If I put up a PR for an iOS codebase I don't know and a code reviewer rejects it because it doesn't use the right framework, or if a designer builds a new feature and it gets rejected because it doesn't follow the right architectural patterns, these are failures of automation.
Every team should be writing the CLAUDE.md's, REVIEW.md's, skills, and docs that enable agents to productively work in their codebase with zero additional context from the prompter. This sounds crazy, and at the same time is a natural extension of the stuff engineers have always done: automate, and encode domain knowledge as infrastructure. As the model gets smarter and as the harness matures, this task becomes easier. In the meantime, it is on every team to look for ways to convert their domain knowledge to infra so that Claude can write code better, so that code review catches issues automatically, and so the next person working on your codebase can contribute more easily.
mager retweeted
Agentic AI adoption is on fire at @Uber, and it's changing the way we build, not just in engineering, but across the entire company.
Today, 99% of our engineers use AI tools. More than 70% of pull requests are attributed to local or cloud agents. And our engineers have built 2,500+ agent skills across the software development lifecycle.
Those numbers are exciting, but they led us to a much bigger question:
How do we bring agentic AI beyond engineering?
Finance. Legal. Operations. Marketing. Customer Support. HR. Procurement.
These functions run on complex workflows that are often manual, highly nuanced, and spread across dozens of systems. You can't automate them effectively by looking at process diagrams or documentation. You have to understand how the work actually gets done.
So we created something called Agentic Pods.
The idea is simple.
We handpicked ~30 of our most AI-proficient engineers (people with deep knowledge of Uber's systems) and paired each of them with a domain expert from a business function.
Then we gave every pod just two weeks.
• Days 1 – 2: Shadow the expert. Observe every step. Document workflows. Ask questions. Build intuition.
• Day 3: Prioritize opportunities based on scale, repetition, business impact, and data availability.
• Days 4 – 5: Build a working agent alongside the person doing the job.
• Days 6 – 9: Validate with several others performing the same work. Does it generalize? Does it actually make their job better?
• Day 10: Ship.
In just the past two months, we've run 16 Agentic Pods across 16 different business functions.
• Capital allocation across 150 cities: 15 hours → 30 minutes.
• Financial pacing reports: 2 days → 10 minutes.
• Marketing web quality assurance: 2 weeks → 50 minutes.
• Support workflow creation: 9,000 manual workflows → self-service automation.
The productivity gains are impressive, but what surprised us most wasn't the speed.
• It was how quickly engineers embedded in unfamiliar domains uncovered opportunities that had been hiding in plain sight.
• The biggest wins rarely come from automating one task. They come from rethinking an entire workflow. Once you redesign the workflow around AI, you often eliminate handoffs, remove unnecessary approvals, replace legacy tooling, reduce vendor spend, and dramatically accelerate decision-making.
• The workflow becomes the unit of automation - not the individual task.
• The most impactful agent skills cut across teams, orgs, functions, tools, and systems.
The biggest lesson? The best AI opportunities are rarely visible from the outside.
You discover them by sitting next to the people doing the work, understanding every friction point, and building with them, not for them.
We're now forming a dedicated team to scale this further and go deeper. They'll deeply understand the work, redesign it from the ground up, and use AI to fundamentally change how the business operates.
It's exciting times!
mager retweeted
Hot yoga reduces depression symptoms.
Eight weeks of yoga in 105°F reduced depression scores by an average of 13 points in adults with moderate depression, and 44% of participants became nearly symptom-free.
There was also a dose-response relationship - each additional session was associated with nearly another 1-point reduction in depression scores.
It's not a standalone replacement for mental health care, but heat + movement is so powerful for physical and mental health.
