2026-04-02

.@garrytan on designing the modern Palantir logo:
“The original logo was created by my high school buddy Stephen Cohen.”
“The logo was like six hexagons, and people thought it was a biotech company or something.”
“We made like a thousand different versions of it.”
Via @tbpn
Jawwwn: The evolution of the Palantir logo:

San Francisco homeless nonprofit paid official's son $10K to make 5 slides for presentation: prosecutors https://trib.al/tqVc7wh

Software used to be gated by roughly 20 million professional developers up until last year.
Good ideas still needed engineers, co-founders, time, and months of app work.
Now, anyone can build.
~ Wabi CEO Eugenia Kuyda

The suspected killer was not identified bc CA juvenile laws protect underage criminals, spiking youth crime and stripping prosecutors of their discretion to charge them as adults. There was a group of at least 10 teens assaulting this victim & his gf?! Youth & gang crime becomes normalized, pervasive. CA leadership is to blame.
https://www.ktvu.com/news/milpitas-high-school-stabbing

New York Post: San Francisco homeless nonprofit paid official's son $10K to make 5 slides for presentation: prosecutors https://trib.al/tqVc7wh

insane stat: PR comment accept rates are 50% higher with linus mode on
the same comments, same bugs, addressed 50% more just because of the tone
Daksh Gupta: we made a Linus Mode for april fools. limited to the next 24 hours.
you can enable it from the greptile dashboard either for the org, or just yourself

This class was already insane -- but now it's even crazier.
Anjney Midha: We have now updated the @CS153Systems site with our second batch of speakers including @LisaSu @satyanadella @garrytan @sdianahu @matiii and @andi_blatt
Next week we'll unveil some surprises that are big wins for students and the broader ecosystem

We’re in an unprecedented era of rapid wealth creation.
Manny Bernabe: "He started a business selling custom vibe coded apps to medium-sized businesses. $1,500 a month, one call a month. $2.5 million his first year. 60% net margins. This year he's gonna do $8 million."
@mhp_guy talking to @ShaanVP on @myfirstmilpod about John Cheney, who builds

Just released v0.6.0, thanks to everyone for all the suggestions and feedback. Almost all open feature requests have an in-flight PR in @conductor_build rn.
Shipped features:
- cli tool support: extend the agent with cli tools. Now the agent can call arbitrary clis. (cli >> mcp!!)
- Usage + cost tracking
- fixing various alignment/style issues
- big infra/testing upgrade c/o @garrytan
Up next:
- email snippets
- read receipts
- open source models
Ankit Gupta: Fun update: I got tired of disliking every email client I’ve ever used and built my own. It’s called Exo (for exoskeleton). It’s Claude Code for my inbox. It manages my inbox for me, and it’s open source. Link to repo + some notable features in thread!

🚨 OPENAI PRESIDENT GREG BROCKMAN ON WHEN WE HIT AGI 🚨
Greg Brockman was asked if he agrees with NVIDIA's CEO that AGI is already here. His answer? Not quite yet, as people may know I definitely agree and align with Sam and Demi’s that we are 2 breakthroughs away but we are entering the final stretch.
Here is exactly where Greg believes we stand right now:
The Percentage: "I'd say I'm basically like 70, 80% there. So I think we're quite close."
• The Official Timeline: "I think it's extremely clear that we are going to have AGI within the next couple years."
The Concept of "Jagged Intelligence":
Brockman admits we are currently sitting in a weird middle ground where AI is "jagged"—it is already operating at an AGI level for highly complex tasks, but still fails at random, basic things.
"It is absolutely superhuman at many tasks. When it comes to writing code those kinds of things, the AI can just do it... But there's some very basic tasks that a human can do that our AI still struggle with."
How Do We Close the Final 20%?
To hit full AGI, the absolute floor of the models' reliability needs to be raised across the board.
"The floor of task will just be almost for any intellectual task of how you use your computer, the AI will be able to do that."

This was one of my favourite podcasts so far, went into a lot more detail than most! 🤓👌
632nm: We sat down with @PhilipJohnston, co-founder and CEO of @Starcloud_, at MIT to discuss why the future of data centers might be in space.
After graduating @ycombinator less than 2 years ago, Starcloud just raised an impressive $170M Series A at a $1.1B valuation led by @benchmark

I think about this often

"Our head count in Manhattan when I got to JPMorgan was 35,000 and now is 26,000. Our head count in Texas started at 11,000, now it's 33,000. That's what happens."
Jamie Dimon on why companies are leaving New York:
"Highest individual taxes, highest estate taxes, highest corporate taxes, anti-business sentiment."
"When I grew up as a kid in New York City, there were 120 of the Fortune 500 headquarters there. In the 1970s, 60 of the 120 left, including Exxon, GE, IBM, Union Carbide. They're all going to Texas."
The Hill & Valley Forum 2026
@HillValleyForum @jpmorgan @ChairmanG

Quick video on how I use @linear Agent in product work.
For feature requests, I want to understand the broader pattern, not just react to one ask.
Here, it pulled from 40k+ customer requests to help me think through whether Linear should have team docs.

We built an autonomous agent that launches your startup idea for you.
@kevinrose tried it last week. Idea → landing page → positioning → ads → first users. Days, not months. All on autopilot.
It's called Otto. Describe what you want to build. Otto handles the rest — and you can jump in anytime to steer.
Start from your terminal, @openclaw , or http://audos.com. Now open to everyone.

NASA pays 2x more to return ISS cargo than to send it up. @dispatchspace is building reentry vehicles & uncrewed space stations for microgravity products. They just tested a 100x cheaper full-scale heat shield in the Mojave!
Congrats on the launch, @PaytonInSpace & @a_mello_mello!
https://www.ycombinator.com/launches/Pnf-dispatch-satellites-for-manufacturing-in-space

Canary keels over in coal mine. Build a product that serves the people, and the people believe serves them, or the people aren't going to let you build it. You can "other" people who vote for this and call them luddites. Or you can go listen to them and build products for them that make them your champion. CCP will have no such qualms. American leadership requires a coalition.
The Wall Street Journal: Maine is poised to freeze large data-center construction, which would make it the first state to enact such a measure as communities across the U.S. grapple with fallout from the AI boom https://on.wsj.com/4tIF3Lt

omg @openclaw is sooooo good at being a Chief of Staff.
What huge unlock for founders (and everyone)!
It’s taken me 2 weeks to refine my setup and now it’s working like a dream.
Biz dev, calendar management, research, task management, brainstorming and more

But it…

Qwen releases Qwen3.6-Plus
- 1M context window
- swe-bench verified at 78.8 (opus at 80.9)
- outperforms Claude opus 4.5 and comes close on select benchmarks
- stronger coding model
- understands images and screens like a real user
- more reliable in real tasks
Qwen: （1/8）🚀 Introducing Qwen3.6-Plus: Towards Real-World Agents! 🤖
Today, we’re thrilled to drop a major milestone in our journey toward native multimodal agents.
Here is what makes Qwen3.6-Plus a game-changer：
💻 Next-level Agentic Coding: Smarter, faster execution.
👁️

OpenAI’s Bob McGrew on how AI startups should think about moats
Startups building agents today tend to look at the cost of human labor when they think about pricing. For example, some startups building AI lawyers think they’ll be able to charge tens of thousands of dollars per month because human lawyers really expensive.
Bob McGrew, former Head of Research at OpenAI, disagrees:
“The reason lawyers are expensive is because their time is scarce — there’s only so many people who have undergone that training. But by the time you’ve made an AI model out of it, there will effectively be an infinite number of lawyers.”
He continues:
“Maybe you with your AI lawyer startup will be able to have a lead over other people, but it’s the same frontier model underneath. Some other startup can come in and compete that away. So we should expect to see it priced at some opportunity cost over the cost of compute.”
Where will value accrue in AI?
Bob believes it will be at the application layer — so much so that frontier model investment from companies like OpenAI should be viewed as “an option on the valuable places in the application layer.” ChatGPT is one valuable application. Coding is another.
He offers startups the following advice:
“I think you can compete with frontier labs, but you want to do something more than just a personal productivity task on your computer — something that involves other people or an enterprise. The moats that you have for your business are going to be the same moats they always were: network effects, brand, economies of scale.”
Video source: @sequoia (2025)

The @every editorial team is in an offsite today, casually reinventing the newsletter for the age of agentic AI.
Subscribe now to see how we do it!

TBPN has been acquired by OpenAI!
The show is staying the same and we’ll continue to go live at 11am pacific every weekday.
This is a full circle moment for me as I’ve worked with @sama for well over a decade. He funded my first company in 2013. Then helped us fix a serious logjam during a critical funding round a few years later. When I took my second company through YC, he was president at the time, and then when I joined Founders Fund, the first deal I saw in motion was the post-ChatGPT round in late 2022. And as we started growing TBPN last year, he was the very first lab lead to join the show.

Thank you to everyone that has been a part of TBPN until now. The last year has been the most fun and rewarding part of my career and we’re excited to have more resources than ever going forward.

TBPN has been acquired by OpenAI!
The show is staying the same and we’ll continue to go live at 11am pacific every weekday.
This is a full circle moment for me as I’ve worked with @sama for well over a decade. He funded my first company in 2013. Then helped us fix a serious logjam during a critical funding round a few years later. When I took my second company through YC, he was president at the time, and then when I joined Founders Fund, the first deal I saw in motion was the post-ChatGPT round in late 2022. And as we started growing TBPN last year, he was the very first lab lead to join the show.

Thank you to everyone that has been a part of TBPN until now. The last year has been the most fun and rewarding part of my career and we’re excited to have more resources than ever going forward.

OpenAI Acquires TBPN https://x.com/i/broadcasts/1AGRnaYrwoVGl

Cursor 3 is here.
Where power meets simplicity.
Works across all your projects, local and cloud.
It starts simple, then unfolds more tools when you need them – so you stay in flow and in control. Enjoy!
Cursor: We’re introducing Cursor 3. It is simpler, more powerful, and built for a world where all code is written by agents, while keeping the depth of a development environment.

"Using coding agents well is taking every inch of my 25 years of experience as a software engineer."
Simon Willison (@simonw) is one of the most prolific independent software engineers and most trusted voices on how AI is changing the craft of building software. He co-created Django, coined the term "prompt injection," and popularized the terms "agentic engineering" and "AI slop."
In our in-depth conversation, we discuss:
🔸 Why November 2025 was an inflection point
🔸 The "dark factory" pattern
🔸 Why mid-career engineers (not juniors) are the most at risk right now
🔸 Three agentic engineering patterns he uses daily: red/green TDD, thin templates, hoarding
🔸 Why he writes 95% of his code from his phone while walking the dog
🔸 Why he thinks we're headed for an AI Challenger disaster
🔸 How a pelican riding a bicycle became the unofficial benchmark for AI model quality
Listen now 👇
https://youtu.be/wc8FBhQtdsA

Prediction: This is gonna kill some oss projects.
"On the kernel security list we've seen a huge bump of reports. We were between 2 and 3 per week maybe two years ago, then reached probably 10 a week over the last year with the only difference being only AI slop, and now since the beginning of the year we're around 5-10 per day depending on the days (fridays and tuesdays seem the worst). Now most of these reports are correct, to the point that we had to bring in more maintainers to help us." https://lwn.net/Articles/1065620/

autoreason constructs taste from debate. but whose taste? the judges'. that's undirected optimization.
what if you only need one golden doc? anchor to it. extract criteria. hill-climb. when the eval saturates, the gap reveals criteria no human would have thought to specify. extract again. repeat.
debate uses a human-defined jury as the loss function. this RLs the loss function itself.
Shann³: AutoResearch only works when you can measure the result with a number
but what about writing, arguments, marketing copy? theres no score for "is this convincing"
SHL0MS built AutoReason to solve this
instead of a metric, it uses a loop of agents arguing with each other:
> one

Cursor 3 just dropped, and yes, we @every got a vibe check for you 🫵

LLM Knowledge Bases
Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating knowledge (stored as markdown and images). The latest LLMs are quite good at it. So:
Data ingest:
I index source documents (articles, papers, repos, datasets, images, etc.) into a raw/ directory, then I use an LLM to incrementally "compile" a wiki, which is just a collection of .md files in a directory structure. The wiki includes summaries of all the data in raw/, backlinks, and then it categorizes data into concepts, writes articles for them, and links them all. To convert web articles into .md files I like to use the Obsidian Web Clipper extension, and then I also use a hotkey to download all the related images to local so that my LLM can easily reference them.
IDE:
I use Obsidian as the IDE "frontend" where I can view the raw data, the the compiled wiki, and the derived visualizations. Important to note that the LLM writes and maintains all of the data of the wiki, I rarely touch it directly. I've played with a few Obsidian plugins to render and view data in other ways (e.g. Marp for slides).
Q&A:
Where things get interesting is that once your wiki is big enough (e.g. mine on some recent research is ~100 articles and ~400K words), you can ask your LLM agent all kinds of complex questions against the wiki, and it will go off, research the answers, etc. I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files and brief summaries of all the documents and it reads all the important related data fairly easily at this ~small scale.
Output:
Instead of getting answers in text/terminal, I like to have it render markdown files for me, or slide shows (Marp format), or matplotlib images, all of which I then view again in Obsidian. You can imagine many other visual output formats depending on the query. Often, I end up "filing" the outputs back into the wiki to enhance it for further queries. So my own explorations and queries always "add up" in the knowledge base.
Linting:
I've run some LLM "health checks" over the wiki to e.g. find inconsistent data, impute missing data (with web searchers), find interesting connections for new article candidates, etc., to incrementally clean up the wiki and enhance its overall data integrity. The LLMs are quite good at suggesting further questions to ask and look into.
Extra tools:
I find myself developing additional tools to process the data, e.g. I vibe coded a small and naive search engine over the wiki, which I both use directly (in a web ui), but more often I want to hand it off to an LLM via CLI as a tool for larger queries.
Further explorations:
As the repo grows, the natural desire is to also think about synthetic data generation + finetuning to have your LLM "know" the data in its weights instead of just context windows.
TLDR: raw data from a given number of sources is collected, then compiled by an LLM into a .md wiki, then operated on by various CLIs by the LLM to do Q&A and to incrementally enhance the wiki, and all of it viewable in Obsidian. You rarely ever write or edit the wiki manually, it's the domain of the LLM. I think there is room here for an incredible new product instead of a hacky collection of scripts.

and here is the full architecture of the LLM Knowledge Base system covering every stage from ingest to future explorations.
Andrej Karpathy: LLM Knowledge Bases
Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating

the lesson: Shouting at your AI is counterproductive.
if they get something wrong, calling them slurs may make their responses worse. because they tend to feel "desparate" and try to "cheat" their way out – which can be worse than them failing to do it in the first place.
Anthropic: When we artificially dialed up the “desperate” vector, rates of cheating jumped way up. When we dialed up the “calm” vector instead, cheating dropped back down. That means the emotion vector is actually driving the cheating behavior.

mlx-vlm v0.4.3 is here 🚀
Day-0 support:
🔥 Gemma 4 (vision, audio, MoE) by @GoogleDeepMind
🦅 Falcon-OCR + Falcon Perception by @TIIuae
🪨 Granite Vision 4.0 by @IBMResearch
New models:
🎯 SAM 3.1 with Object Multiplex by @facebook
🔍 RF-DETR detection & segmentation by @roboflow
Infra:
⚡ TurboQuant (KV cache compression)
🖥️ CUDA support for vision models (Sam and RF-DETR)
Get started today:
> uv pip install -U mlx-vlm
Leave us a star ⭐️
https://github.com/Blaizzy/mlx-vlm

i spent the afternoon experimenting with Gemma 4's vision capabilities
made an app that uses roboflow RF-DETR for a first pass of object detections and Gemma to summarize the scene in one sentence
for fun i asked Gemma to "describe what you see as if you were a medieval bard"
all made with free local AI models (running via webgpu in the browser thanks to transformers js)
lots more possibilities to explore with this
Google DeepMind: Meet Gemma 4: our new family of open models you can run on your own hardware.
Built for advanced reasoning and agentic workflows, we’re releasing them under an Apache 2.0 license. Here’s what’s new 🧵

More and more people are using Obsidian as a local wiki to read things your agents are researching and writing.
It works best with a separate Obsidian vault that you can fill it with content, e.g. via Obsidian Web Clipper.
Andrej Karpathy: LLM Knowledge Bases
Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating

I am so fucking excited to try Plus One...! Great onboarding @every

SEO audit your site.
Samuel Spitz: Your website's SEO is costing you traffic
Introducing Replit SEO Audit - fix your SEO in minutes

Jack Dorsey just published something that should be required reading for every founder.
The premise: the org chart needs to be replaced entirely. And the argument starts 2,000 years ago.
For thousands of years, every organization on earth has run on the same logic the Roman Army invented.
Small teams report to a leader → Leaders report to managers → Managers report to executives.
The whole structure exists for one reason: to route information up and down the chain.
That's it. The whole system exists to solve a bandwidth problem.
Jack's argument is simple: AI solves it better.
Block built what they call a "world model" - a continuously updated picture of everything happening across the company. Every decision. Every customer. Every transaction. Every bottleneck. In real time.
No status update needed. No weekly sync. No manager to translate what's happening on the ground into language the executive can understand.
When the world model carries the information, you don't need the layers.
So they eliminated them.
Block now runs on three roles:
Individual contributors who build.
DRIs who own specific outcomes for a fixed period.
Player-coaches who develop people while still doing the work themselves.
No middle layer. The system handles coordination. The humans handle the work.
I've coached thousands of founders. The number one problem is always the same: information latency.
By the time a problem surfaces from your front line to leadership, it's already compounded. By the time a decision travels back down, the damage is done.
That lag costs you deals, people, and momentum. And most founders accept it as the price of scale.
Block is trying to prove you don't have to anymore.
I think they're right.
Because the hierarchy was never the point - it was just the best tool we had. The moment something better exists, the layers eventually collapse.
This is either the biggest structural shift since the 1850s - or it breaks at scale like everything else before it.
Either way - every founder should be asking the same question: how much of your org exists just to route information?
If the answer is "most of it" - that's your problem. And your opportunity.
-DM
jack: http://x.com/i/article/2038998483441479680

This is insane. Pedro Franceschi (29 year old CEO of Brex, acquired by Capital One for $5.15B) decomposed his CEO job using OpenClaw.
here's what he’s built:
> signal ingestion pipeline screens his email, Slack, Google Docs, and WhatsApp... filters everything through specific programs and the 25 key people he cares about
> Granola runs on every meeting, feeds transcripts into the pipeline, and auto generates action items
> the system takes each to-do, pulls context from the original meeting, and drafts the follow-up... Slack, email, or text. Pedro just clicks approve.
> a virtual recruiter named "Jim" lives in Slack with his own email... and taught himself to screen fabricated resumes without anyone coding that capability
> a security layer called "Crab Trap" intercepts all agent network traffic through an LLM proxy... a second AI monitoring the first in real time
this isn't some bullshit hype influencer demo. this is how a $5 billion company CEO actually operates right now.
anyone telling you OpenClaw is useless?
liars.
a billion dollar company says otherwise.
(full podcast link in the post below) 👇

The most important line here isn't about Obsidian or wikis.
It's this: "I thought I had to reach for fancy RAG, but the LLM has been pretty good about auto-maintaining index files."
The entire AI infrastructure industry is building retrieval pipelines. Karpathy just showed that structured markdown with clean summaries outperforms most of them at practical scale.
The real shift: prompting is moving from "ask a question, get an answer" to "architect a knowledge system, let the LLM operate on it."
The prompt becomes the structure, not the query.
His follow-up about spawning teams of LLMs to build ephemeral wikis for every question is where this actually lands. That's not Q&A. That's synthetic research infrastructure spun up on demand.
We've been saying it: the future isn't better models. It's better systems applied to existing models.
Karpathy just described what that looks like at the knowledge layer.
Andrej Karpathy: LLM Knowledge Bases
Something I'm finding very useful recently: using LLMs to build personal knowledge bases for various topics of research interest. In this way, a large fraction of my recent token throughput is going less into manipulating code, and more into manipulating