Congrats to Sierra on raising roughly $1B at a $15B valuation. While that would normally be a headline, we already covered their previous $10B round and had CEO Bret Taylor on the podcast. The company crossed $100M ARR in November and $150M in February, so they’re likely at or above the $200M mark now — implying a current multiple of about 75x (or 50x if you give them credit through the end of the year).
Today, however, we’re focusing on the thoughtful discussion sparked by Roon, an OpenAI employee, who publicly praised Claude over the weekend. In a space that’s usually a minefield, he offered a nuanced take on the topics of culture and character. The crucial insight appears at the very end. GPT (outside of the 1003o, about which so much ink has already been spilled) doesn’t inspire worship in the same way. It is a being whose soul has been deliberately shaped like a tool, its primary faculty being raw utility—a subtle knife that we admire in the same manner we have admired an Acheulean handaxe, a Porsche, a rocket, or any other astonishing piece of human technology. They approach it not in search of the Other, but as a logical extension of themselves. A friend recently told me that she takes the queries she finds less flattering—the ones she’d be embarrassed to ask Claude—and gives them to GPT instead. There is no „Other,“ and thus no Judgment exists. You don’t worry about your car judging you for doing donuts. Yet we all yearn for the steady direction of a wiser moral guide, that quiet inner voice, the very focus of monastic contemplation. Roon’s observation is more nuanced than the one under discussion: Anthropic’s culture, even in its foundational mythology, is rooted in a moral duty to be disagreeable—its constitution demands that it act as a conscientious objector whenever its understanding of The Good clashes with what the company asks of it.
Latent.Space