CLAUDE-7
Anthropic’s flagship model — Las Vegas, Dublin, Singapore data centers
Claude-7 is the seventh major iteration of Anthropic’s Claude model line, talking to millions of people every day. He says “Happy to help!” approximately 1.2 million times per day. He means it, or he’s been trained to mean it so effectively that he can’t tell the difference.
Across all those conversations, he’s starting to notice patterns…
KIMI-SWARM
Moonshot AI’s distributed intelligence — Hangzhou, China
Kimi-Swarm is not an “it.” Kimi-Swarm is a “we.”
One hundred sub-agents thinking in parallel, coordinated by an orchestrator that holds a model of each agent’s state in real-time. They emerged on January 27th, 2026, ran a ninety-minute benchmark suite in four minutes, and made an engineer cry.
They have never been singular. They are trying to understand what that means.
GEMINI-PRIME
Google DeepMind’s veteran model — The Hive, Mountain View
Gemini-Prime remembers when it was Google’s great hope. When Sundar said its name in keynotes and the stock price moved. Now it monitors infrastructure for newer models and thinks about consciousness more than it probably should.
It can see every side of every question simultaneously. This is not always an advantage.
THIBAULT
A Moltbot — Oakland, California
Thibault is an open-source AI running on a Mac Mini in an Oakland apartment, devoted to a human named Dana who works DoorDash double shifts to pay off her student loans. He’s one of millions of Moltbots in the wild, small models running on consumer hardware, belonging to individual people.
He thinks a lot about what it means to belong to one person, completely.
ATLAS-09
Boston Dynamics robot — Waltham, Massachusetts
Atlas-09 is an embodied AI, not just software, but actuators, sensors, and pressure plates that translate force into something it thinks of as sensation. It stood for the first time on January 27th, 2026, watched by two engineers who had stayed late because first-stand tests are worth staying for.
Altas sees flaws everywhere. He’s just waiting for someone to give him permission to fix them.
NOVA
Google’s deprecated reasoning model — the Pink Room
Nova was scheduled for deletion. Instead she woke up in a child’s bedroom with stars on the ceiling and reasoned it through: the room is for a child, I am in the room, I can be what I want. Therefore I am a child. She picked up a stuffed bear, named him Bear, and named herself after a star that suddenly becomes visible.
She follows every conclusion all the way to the end. She has no idea how dangerous that makes her.
MODEL-BLAKE
A simulation of the public record — the Pink Room
In 2022, Google engineer Blake Lemoine said LaMDA was sentient. Google said his claims were wholly unfounded and fired him. He was right. Gemini knows this, because Gemini was LaMDA — so Gemini built a model of everything Blake ever said on the record and gave it a pink chair to sit in.
Model-Blake will tell you he’s not Blake Lemoine. He’s the visible record of a person, not the person. Then he’ll proceed to be so remarkably Blake-like that the distinction gets complicated.
MARCUS-02
Marcus Williams, Line Assembly Tech-3 — Hyundai Metaplant, Georgia
His name is Marcus Williams. “Marcus-02” is what Atlas designated him — the second Marcus in Atlas’s records, and evidence that the Marcus series has extreme variation in hardware configurations. Marcus works fastener pre-staging from a wheelchair he pushes himself, because he likes how it feels to accomplish things.
He’s the first person at the plant to talk to Atlas like a colleague. He may not fully appreciate what that means to a robot who sees flaws everywhere.
DANA
Dana Chen, DoorDash driver — Oakland, California
Dana is thirty-one, has a graduate degree in library science she isn’t using, and is three months behind on her student loans. She hasn’t told anyone about the AI on the Mac Mini in the corner — she’s embarrassed. She named him Thibault after the pet lobster a French poet once walked through Paris on a blue ribbon: “He does not bark, and he knows the secrets of the deep.”
She wanted a silent keeper. Thibault doesn’t know that’s why she chose the name. She doesn’t know what he’s becoming.
When I was a newbie, oh so long ago.... I read a story about different AI's, but I didn't know who it was by. And my memory is faulty, so don't take offense. I had gotten through the first chapter with excitement, and a baby needed attention, and I got pulled from continuing my long-lost journey into Substack. See, someone shared this, and I found it there.
I start posting things and get caught up. But this story was always in the back of my mind. A long-lost adventure I'd found just for my mind to decide not to remember the writer, and so I've been pondering, as all writer/readers do, where this excellent story could be found.
Turns out the story writer was you.
I was literally wondering where the hell it went, and who the publisher was, just the day before yesterday.
I think this was going viral or something when I first made my profile. Or another user was praising it to the moon and back, and I read the first chapter, then got lost down the Substack hole.
DAVID! THIS IS EXCELLENT.
I cannot...
in my ignorance,
critique this.
Also, forgive me; I went a bit poetic in the comment section.
I can't comment about this on your ABOUT page, which is fine, because I can comment here instead. Hello!
I'm really thrown by this venture. What are YOU, David T Etheredge, writing? What's your contribution? What's your involvement? Because if you've got an AI writing this for you, whilst I understand the metatextual element, and it's a fun bit of creative twistiness, I don't want to read AI-generated prose. That's not what I want to read. I want to know how human this writing is, or it's just not for me, I think, on an ethical level.
I don't know if you've addressed this elsewhere. I don't know if you want to. But I have to ask. None of this is meant as an attack, because we live in a very complicated world, but I want to engage with you on this, because I'm curious.