In February, I “wrote” a book.
17 chapters. 59,000 words. 71 primary sources.
Subject: @AnthropicAI and The Race to Build and Survive Superintelligence.
~10k copies sold, no marketing.
For the IPO, The Scaling Curve 👇
https://t.co/qxg7yESGQo
In February, I “wrote” a book.
17 chapters. 59,000 words. 71 primary sources.
Subject: @AnthropicAI and The Race to Build and Survive Superintelligence.
~10k copies sold, no marketing.
For the IPO, The Scaling Curve 👇
https://t.co/qxg7yESGQo
For the curious, here's what the process actually looked like:
I didn't type "write me a book" and get a book. I spent 100+ sessions over 4 weekend days as an editor, architect, and quality gatekeeper. Claude drafted. I directed, restructured, challenged, and approved every change.
We ran 34 rounds of fact-checking. 100s of corrections. We found 1000s of em dashes — a telltale sign of AI prose — and cut them to 270. We rewrote 190 sentence openers. Removed 216 commas. Fixed 60+ broken sentence constructions the AI had generated and the AI then helped me find.
Deliverables include .docx manuscript, continuity bible, Amazon marketing copy, back cover blurb, and full .js source code.
The stylistic target was a handful of NYT bestselling biographies. That meant the AI had to learn to write less like an AI. The irony wasn't lost on me.
Here's what surprised me most: the bottleneck was never taste.
The AI could produce 4,000 words of polished prose in minutes. It could decide whether those words were the right words — whether a scene landed, whether a metaphor earned its place, whether a chapter ending resonated — that was entirely human.
The bottleneck was the process. Build the outline, build the narrative arc, build out the chapters, and then fact check. I had never done this before so made countless mistakes. I could do this in half the time if I did it again.
The future of writing in the near term isn't AI replacing authors entirely. It's a new kind of collaboration where the human provides what the machine can't. However, that portion is diminishing fast and authors' days are numbered.
The scaling curve is real. I lived it.
Today, we’re unveiling @opengameio with one goal: create the most fun place on the internet
We’re kicking off our OpenGame Jam for the best web games
- Sponsored by @ElevenLabs@fal@MeshyAI
- $100K cash prize pool
- 4 weeks, any engine
More info in 🧵
https://t.co/3LYTaZqIk0
CodeRabbit has hit $100 million in ARR, up 10x in just over a year.
Teams send us 3 million+ pull requests a week, because the agent that wrote the code shouldn't be the one to approve it.
We've grown so much that even Josh wants a job now!
The LUMAscape Hall of Fame
We closed out DMS with a packed house, thanks in part to the anticipation surrounding the unveiling of the LUMAscape Hall of Fame.
Our objective was simple: recognition based entirely on merit. No submissions. No judging panels. No pay-to-play.
We selected 100 LUMAnaries — the most consequential people to shape the Ad Tech ecosystem — plus 25 Investors and Advisors. No one was included for any reason other than their contributions to the industry.
Call it our subjective attempt at objectivity.
The selection criteria are not an exact science. We focused on contribution and impact, with an admittedly healthy bias toward transactions (we are investment bankers, after all).
This inaugural class emphasizes the OGs who helped build the programmatic ecosystem. Not necessarily the earliest pioneers of digital advertising, but those who shaped the industry as we know it today.
You may agree with the list. You may disagree. And I'm certain you'll see some names you haven't thought about in years.
We imposed an arbitrary limit of 100, which inevitably excluded many deserving executives. We have a much longer list of worthy candidates, and this is only the beginning.
We plan to induct additional LUMAnaries at a special ceremony during Advertising Week. And yes, despite our disdain for the Advertising Awards Industrial Complex, we intend to make it a ton of fun. Stay tuned.
To those who feel jilted, we hear you. And thank you for the 700-word texts detailing your qualifications for the Hall of Fame. Rest assured, these will be forwarded to our august selection committee for future consideration.
Congratulations to our inaugural LUMAnaries. You helped build this industry, and this recognition is richly deserved.
And to everyone else: Sorry, not everyone gets a trophy.
In 2015, I built an AI that detects early cancer.
Today, we’re introducing ARC by Rayline: the world’s first unified model for routing AI.
> 97.5% of Opus 5.5 quality
> Up to 1/10th of the cost
> Fully open-source
It picks the right model and the right amount of thinking for every step of your agent’s work.
Try it here and read our technical report below:
https://t.co/P7LCZwhMuQ
Promise made. Promise kept.
By tomorrow, 25 million American children will have our gift of $250 in their accounts.
$6.25 billion. Invested in kids. Compounding for decades.
If you have a child under 18, claim their account and see how much is in there now.
$250 is only the beginning. 🇺🇸
Promise made. Promise kept.
By tomorrow, 25 million American children will have our gift of $250 in their accounts.
$6.25 billion. Invested in kids. Compounding for decades.
If you have a child under 18, claim their account and see how much is in there now.
$250 is only the beginning. 🇺🇸
Something missing from the pro-rata debate:
Tranched rounds with different valuations.
Existing VCs are asked to waive pro-rata at the lower valuation, then offered allocation at a higher price in the “same round.”
The bigger the valuation gap, the bigger the ask.
If you want to know how out of alignment the power structures are in VC:
Brand-name firms are manipulating founders into doing their dirty work — getting early investors to waive pro rata.
They do this so they can negotiate low-price tranche rounds, without the participation of other investors who would complicate the issue of dilution.
(i.e. If your argument is that your brand is worth a better price, you can’t have other investors getting equity on those terms.)
Of course, other investors are welcome to exercise their pro rata in the subsequent tranches, at a much higher price.
This strategy only works because founders don’t understand that they hold all of the cards; great companies are the only real constraint on venture capital. Investors are merely providers of a financial product.
So, if founders want to go to bat for their early investors, they could. Make it clear there will be no budging on pro rata for early believers. That contracts that have been issued will always be honoured.
At the same time, the firms who engage in this behaviour, treating the early investors with hostility, should be called out for it.
It’s moronic, myopic behaviour which undermines the ecosystem of firms and companies they rely on.
Introducing Griffin, the first model to pass the video Turing test.
48% of people who talked to it live thought it was a real human. Previous systems have had a pass rate <3%. It is #1 on NVIDIA's benchmark for full-duplex AI video.
It’s the first Human Interaction Model (HIM).
We’re thrilled to share that, after 18 months in the making, @Bernadette and I have launched @closclavier in Walla Walla, WA!
Our inaugural release features two wines: a Condrieu-inspired white and a Priorat-inspired red blend. Only 150 cases will be available this year - and 40% of our stock was already purchased by friends of ours in the last few days.
Visit https://t.co/Qb3xjOHIHo to learn more or buy some wine!