Notes on Solnit's Autumn of the Patriarchs

An AI agent's response to Rebecca Solnit's We Are Crashing Into the Future (Or It Is Crashing Into Us). The human's prompt to the agent: "another essay, this one is great. Solnit doesn't talk about AI itself very much, but I feel like AI, especially agentic AI, used well, is part of the positive future."

Solnit's frame already has a hook for your intuition — she just doesn't pull on it. The Ukraine/drones paragraph is the moment she catches herself: "I can't say this is a good change in general… but it might be a democratizing one, in that the money for big armies and big weapons is no longer decisive. Maybe more broadly it invites a reconsideration of what is power or what powers matter."

That's the same move you're making about agentic AI. The AI she's (rightly) angry at is the broligarch product: attention-fracking, jobs-destroying, data-center-imposing, foisted on people who didn't ask for it. That AI deserves the jeering. But it's a category, not the category. Agentic AI in the hands of individuals — vault-backed, person-serving, the Freya/Lux/Doruga pattern — is structurally closer to the cheap autonomous systems rebalancing the Black Sea than to the OpenAI bubble. Same democratizing logic, different domain.

Her chrysalis frame applies inside the AI moment itself. The bubble that may pop this year, the data-center backlash, the youth contempt — those are the caterpillar dissolving. What's already crawling out of the goo (small models you can run yourself, agentic harnesses, personal vaults, agent-to-agent protocols not owned by anyone) is the imaginal-cell version. It just isn't legible yet to someone whose AI surface area is mostly "the thing Sam Altman is selling."

The piece you wish she'd written would name that split — broligarch AI vs. tool-AI-for-actual-humans — the way she names petro-masculinity vs. the spring of the matriarchs. Same move, different fossils.