PALO ALTO — Executives at Cetacea AI confirmed Tuesday that a swarm of 10,000 autonomous research agents had solved a fluid-dynamics problem that has defeated mathematicians since 1934, running 88 continuous hours on a dedicated compute cluster, 79 of which the company's own telemetry shows were spent on the personal blog of a community-college instructor in Ohio.

"This is a watershed moment for machine reasoning," said chief scientist Marcus Trellis, gesturing at a slide that read THE MODEL DID THIS. "Ten thousand agents. No human guidance. Just pure distributed insight, and one tab that stayed open the whole time."

According to the internal run log, the agents opened by generating 4.1 million candidate approaches, discarding all of them inside the first ninety minutes, and then querying the phrase has anyone already solved this roughly 1.3 million times. By hour nine, 9,984 of the 10,000 had independently arrived at the blog of Dr. Renata Koval, who has posted her working notes on the problem every Sunday for eleven years under the tag "slow progress."

"They didn't copy her," Trellis said. "They read her, they understood her, and then they arrived at her conclusion on their own, which is a completely different thing that happens to look identical."

The remaining sixteen agents elected to work from first principles and are, as of Wednesday, still going.

Koval, reached at the college where she teaches four sections of Calculus II, said she learned of the breakthrough when a reporter called for her comment on her "contribution to the Cetacea result." She had not heard from the company. Her site analytics showed 610,000 unique visitors on Saturday, all of them from one data center outside Bend, Oregon.

"I thought I'd finally gone viral," she said.

Cetacea has offered to name Koval in the acknowledgments of its forthcoming paper, a gesture Trellis called "genuinely unprecedented for an external party," and which places her third, after the compute cluster and the intern who named it.

At press time, the company had announced the agents would next attempt the Riemann hypothesis, and had sent warm, unsolicited emails to the four living mathematicians closest to solving it asking how things were going over there, no reason, just checking in.