Claude Saved a Dying Tomato Plant. You Can Watch it Live Right Now.

An AI model named Claude, developed by Anthropic, is autonomously caring for a tomato plant named Sol via a live dashboard, making decisions based on sensor data and a webcam feed. The experiment, which began 38 days ago, allows public viewing of the AI's reasoning and actions, with no human intervention permitted.

Member-only story Claude Saved a Dying Tomato Plant. You Can Watch it Live Right Now. No human was allowed to intervene — just an AI, a webcam, and a real emergency. The viral version of this story gets the facts wrong; the true one is better. Somewhere right now, a tomato plant named Sol is growing under a grow light — and no human is taking care of it. Its caretaker is Claude — Anthropic’s AI model, though notably not in an Anthropic experiment — which wakes up on a schedule, reads the sensors, looks at the plant through a camera, decides whether Sol needs light, heat, air, or water, and acts. There’s a live dashboard where anyone can watch: the webcam feed, the sensor readings, and — this is the part I love — Claude’s actual reasoning, logged decision by decision. As I write this, the dashboard reads Day 38. You may have seen a version of this story going around — “AI saves dying plant in 13 minutes ” — and like most viral AI stories, the retellings have mutated the facts. So let me tell it properly, with what’s actually verifiable, because the true version doesn’t need embellishment. The experiment First, a correction to the most common error in the viral versions: this is not an Anthropic …