Over the course of 21 hours, approximately 950 artificial intelligence agents worked in parallel to investigate more than 200,000 enzymes associated with genetic mechanisms found in nature. They searched for patterns, compared sequences, and explored possibilities that could shed light on biological systems that are still poorly understood. In the midst of this research, something unexpected emerged: a set of genetic sequences with an organization that, in some respects, resembled CRISPR—the system that revolutionized our ability to edit DNA.

The discovered tool was named "ART," but there was a problem: no one knew exactly what this system did. This is precisely where the story gets interesting, because the result did not come from a traditional, step-by-step experiment conducted by a team of researchers; rather, it emerged from a large-scale investigation in which hundreds of AI agents explored various paths until they identified a candidate worthy of scientific attention.

For the first time, we are beginning to see more clearly what happens when artificial intelligence moves beyond simply answering questions about biology and begins to help formulate new hypotheses regarding the mechanisms of life itself. What caught the researchers' attention was not an entirely new enzyme—it was a reverse transcriptase (RT) found in a type of bacterial virus known as a bacteriophage—but rather what existed around it.

One of the Cloud agents examined the DNA near that enzyme's gene and found a long sequence of repeats organized at regular intervals; alongside it lay a gene associated with another protein of unknown function. In isolation, these elements might have seemed like just a few more sequences among billions of pieces of genetic information, but together they formed an unusual architecture. The agent began counting the repeats, measuring the intervals between them, comparing the arrangement to known systems, and searching for potential descriptions within the scientific literature.