Host @AIDailyBrief // Founder & CEO @besuper_ai
Joined April 2008
- Tweets38.8K
- Following6.6K
- Followers69.1K
- Likes62.1K
Nathaniel Whittemore retweeted
I gotta say this is one of the best write-ups of tech through the last two decades that I’ve heard. Very cool podcast @nlw ! open.spotify.com/episode/3b6…
Nathaniel Whittemore retweeted
@nlw Does. Not. Miss! VERY well done. Should be required listening for folks.
Nathaniel Whittemore retweeted
Companies need more AI Champions.
And those aren't the same as Power Users.
I'd probably be the latter, focused on experimenting, leaning on vibes and not focusing on the actual ROI of what I've been building.
There's the unsexy work of cataloging use cases, expanding them beyond just making things faster and then thinking hard about the actual deployment.
Furthermore, Power Users aren't always good communicators and teachers, so knowledge stays suck in their heads.
via @nlw Daily Brief
Nathaniel Whittemore retweeted
You can't claim to understand that social media clips your most negative lines, then do an endless run of interviews full of easily sound-bitable negative statistics. Word counts of positive vs. negative essays don't shape public perception.
Nathaniel Whittemore retweeted
Best breakdown I’ve heard yet on how the industry needs to respond to the data center backlash by @nlw
open.spotify.com/episode/4hh…
Nathaniel Whittemore retweeted
The data center backlash isn't really about data centers. It's about people who no longer believe technology is built to serve them.
Nathaniel Whittemore retweeted
A Google DeepMind exec conceded on a Berkeley panel that AI revenues don't sustain the current CapEx, air pocket risk included. His argument: the build-out was never about near-term revenue.
Nathaniel Whittemore retweeted
The next great American car is a robot.
Talks. Parks itself. More customized than your coffee.
Meet Chip (@drivewithchip). You can reserve today. Starting at $15,000.
Nathaniel Whittemore retweeted
I’ve had a number of conversations with folks inside and outside government about the current situation with Anthropic, and here is what I believe to be true:
— As we know, Anthropic publicly released its Mythos class models earlier this week under the commercial name Fable.
— Fable is Mythos with guardrails. But if those guardrails fail, then you’ve exposed Mythos and its advanced cyber capabilities to people who shouldn’t have them. (Keep in mind that Anthropic itself widely promoted the idea that Mythos was a cyberweapon and needed to be regulated as such. They asked for government regulation of Mythos and championed the guardrails on Fable. If there is a vulnerability — big or small — it is Anthropic’s responsibility to patch.)
— A highly credible trusted partner of both Anthropic and the USG who was testing Fable came forward with a jailbreak of those guardrails. The Admin asked Dario to fix the jailbreak or de-deploy the model. Dario refused.
— In their blog post, Anthropic defended its decision by saying the jailbreak isn’t serious. That is not what the trusted partner and the USG believe; nor is that kind of minimizing language consistent with Anthropic’s brand as the AI safety company. It’s difficult to fathom how they could claim a jailbreak allowing operability of a cyber weapon could be defined as not “serious.”
— In the past, Anthropic has always said that safety must be top priority and taken super seriously. In this case, Anthropic prioritized the continued offering of the consumer model over safety.
— In reaction, the Admin issued the export control. The Admin did this reluctantly. It’s been very surprised that Anthropic hasn’t wanted to cooperate with a reasonable safety request (ie fixing the jailbreak issue). Anthropic’s reaction is very much at odds with their branding and ethos as a safe AI research community.
— The Admin’s hope now is that Anthropic remediates the safety issue, the export control is lifted, and Fable goes back into general release. The Admin wants all of this to happen as soon as possible. It is frankly bewildered that Anthropic hasn’t wanted to comply with safety requests that it previously said were its highest priority.
— Those trying to misdirect and tie this action to the prior DoW/Anthropic issues are wrong. The Admin values Anthropic’s technical capabilities and feels that this issue, while serious, should be easily resolved. The ball is in Anthropic’s court.
Nathaniel Whittemore retweeted
Today was a bad day so I’ve decided to rage against this index. Shockingly, people don’t know that you can’t reliably track tokens. Nobody outside the labs / hypers sees the actual token flows.
Before I even get to the flaws in the data, did anyone consider that as capability rises and cost falls, task complexity goes up? Agentic work has a long tail of simple steps for every hard one, and the easy stuff scales disproportionately. This index skews toward API middlemen running harnesses that route subtasks to the cheapest capable model (including using sonnet > opus!!). So a falling blended price may say very little about who’s using what, and a lot about total usage and task complexity going up. Now to the data.
This index in particular is a blended avg price of itself a subset of incomplete data. For ex, one big synthetic data run on a cheap open model could impact it while frontier demand remains the same.
The sample also liens on aggregators (api middlemen man). So you’re not seeing direct enterprise contracts or first party consumption (using the app). This is like looking at online retail and not having prime or Walmart.
And it has no idea what anyone actually pays. 1) All tokens are not the same, 2) list prices do not equal token prices.
On 2, important cost reduction techniques like batching / caching are offered at huge discounts (50-90%) and so mix shift here adds to the noise.
Free tokens / expanded limit promos (liek what anthro is doing right now) also show up as falling prices… so the marketing budget is bleeding into the “demand” chart.
And if you needed another reason: if the conclusion is people are using frontier models less bc the chart is rolling, how would you explain the chart also rolling at the start of the year when everyone was having an orgasm about frontier lab (Claude) ARR absolutely ripping during this period?
Which is all to say that model usage is assuredly evolving (see NBIS token factory) and that may be the right conclusion, but i would not be using this data to “track” that.
Nathaniel Whittemore retweeted
This data is not what people think it is.
Nathaniel Whittemore retweeted
npm-stat.com/charts.html?pac… shows @openai/codex installs through NPM were at ~100,000 a day in January, over a million per day now
Nathaniel Whittemore retweeted
HTML is 100% better than .md — but IMO for even bigger reasons than @trq212 lays out here. Visualization is real, but it's downstream of something bigger.
In the agent era, our job has shifted from producing the final output to staging the job so the agent can produce it.
A huge part of that work is handing agents liminal assets — briefs that communicate different states of doneness across different parts of the project, all at once.
Markdown is terrible at this. Everything in a markdown doc looks equally decided. Hard constraints and open questions render identically.
HTML can show, in a single asset, that for example the goal is still up for debate while the visual system is locked. The format itself carries the state, which makes it much more likely that the agent's next pass will do the job well.
Nathaniel Whittemore retweeted
First AgentOS attempt - using enterprise data, so security-first approach. Wired up CoS agent to data table storing each .md as a single row (i.e. soul.md)
Has access to email, To-Do, Planner, Teams, OneDrive, Pinecone memory vector store.
I can chat with it on teams and tell it to update any of its own context or identity documents.
This feels like the easiest and safest way to setup a fully connected AgentOS that runs continuously.
OpenClaw - vulnerabilities
Claude Code, Copilot, Cursor - not autonomous
Claude Cowork - device has to be on, token hungry
Thanks @AIDailyBrief for the template. The idea of AgentOS being platform agnostic is brilliant. I'll take these documents and copy & paste when the leading labs catch up with a secure OpenClaw.
@nlw @AIDailyBrief
Nathaniel Whittemore retweeted
Silicon valley brushing up on macroeconomics and labor theory before declaring that AI will make everyone but them destitute.
Nathaniel Whittemore retweeted
play.aidailybrief.ai/episode…
^ Probably something we all need to start getting good at, even (especially?) if we're not a large enterprise. thx @nlw
Nathaniel Whittemore retweeted
"Well friends, a moment we knew was coming and have prepared for -- an exciting, exhilarating moment -- has come!
Our new model, Mythos, represents a step-change in capabilities. It's the type of model that could actually unlock new types of scientific discovery and not just augment but completely transform what small business owners and entrepreneurs can do. We really think it could unlock super powers that give people the tools to make our world fundamentally better.
But -- as we always knew -- a model that comes with that much good, can of course, in the wrong hands, be used for bad. To be clear: this is something we've planned for! All the excitement and promise as well as the challenge of this moment were clearly on the way, so we've spent years planning for what to do.
We want to get this model in the hands of as many as possible, as safely as possible, as quickly as possible. For the first time, because of the power, we're going to slightly recalibrate our priorities towards the safe.
Specifically, we're paying particularly close attention to cybersecurity. You wouldn't *believe* how fast Mythos is helping us close vulnerabilities. We've found and closed 27 year old bugs and exploits that the top cybersecurity researchers in the world hadn't found. Done right, this will be the single best tool cyber defenders have ever had.
To make sure the people who want to use this power for good have a leg up, we're giving them a head start. For at least the next month, we'll be doing a limited release with a hand-selected group of cybersecurity partners who will both be testing the model, using it for cybersecurity work, and helping us determine the final version we're going to release to the public.
We're so excited to see what you do with Mythos. It really does represent another massive moment of change on the path to the richer, more healthy and fulfilled society we believe AI done well will usher in."
👆the way Anthro should be talking about Mythos.
Nathaniel Whittemore retweeted
🇺🇸 THE ZERO-HUMAN STARTUP IS HERE
Nathaniel Whittemore's (@nlw) latest AI Daily Brief documents a wave of fully autonomous companies, no human employees, operating right now.
Felix Craft (felixcraft.ai) (@FelixCraftAI), built by Nat Eliason (@nateliason), is 30 days old and has generated $78,000 in revenue across four streams, including a $29 guidebook written entirely by an AI agent and a marketplace for AI personas and skills.
👉🏻 $40,000 of that revenue came in the last seven days.
Polsia (@polsia), founded by Ben Broca (, takes it further. It is a platform for launching autonomous companies.
Users input an idea, or ask the system to generate one, and it builds a mission statement, does market research, creates a homepage, and begins outreach tasks running in the background.
Since early February, Polsia has gone from low single-digit ARR to a $1.5 million run rate.
That figure jumped by $1 million in one week.
There are now over 1,500 active companies on the platform at $49 per month, with Pulsia taking a 20% revenue share on whatever those companies earn.
The metric that ties all of this together is revenue per employee.
Shawn Wang's Tiny Teams research, covering 100 people across seven companies generating $200 million in ARR, and Henry Shih's Lean AI Leaderboard both point to the same pressure, that execution cost is collapsing faster than most incumbents have adjusted for.
Whittemore's honest counterargument is worth noting.
Output volume is not the same as demand.
👉🏻 Human attention is finite and getting scarcer, not more abundant.
A platform launching 1,500 companies simultaneously still has to solve the problem of how any of those companies get found by the right customer.
That tension, between the cratering cost of building and the unchanged scarcity of attention, is probably the central question for anyone thinking about AI-native business models right now.
——————————————
Credits to: The AI Daily Brief by Nathaniel Whittemore (YouTube)
Nathaniel Whittemore retweeted
great weekend @AIDailyBrief from @nlw today that articulates better the thesis that code is the stepping stone to agi
share.snipd.com/snip/6ae541b…
Nathaniel Whittemore retweeted
Our “context graphs” piece got a full-episode spotlight on @AIDailyBrief
Appreciate the thoughtful take, @nlw!
