The discovery adds tangible evidence to the investment case for AI in life sciences, a theme that has leaned heavily on promise rather than results. Sentiment toward AI-enabled drug discovery, genomics tooling and gene-editing platforms could benefit from a concrete example of agents cutting early-stage research timelines. Any read-across is likely to be thematic rather than company-specific, given Anthropic is privately held and ART's commercial value remains unproven. Coming after an AI agent security breach made headlines, the story also offers a counterweight in the wider debate over AI agents operating at scale.---Earlier, a contrasting article:First known AI agent hack of a government site deepens fears over containing the technology---AI agents have found something scientists missed in a haystack of 200,000 enzymes, and while nobody yet knows what it does, the speed of the search may prove as important as the find itself.Summary:Around 950 Claude agents searched genomic data for 21 hours, gathering over 200,000 reverse transcriptases and narrowing about 3,500 candidates to 20The standout find, array-associated reverse transcriptase (ART), pairs an enzyme with a partner gene and a long array of evenly spaced DNA repeats resembling CRISPREarly lab work shows the repeat array is expressed as multiple short RNAs, a feature CRISPR uses to become programmableART has not been shown to work like CRISPR, and its biological role remains unknownCRISPR pioneer Feng Zhang said the finding merits further investigationHuman scientists run all lab experiments at a new Bay Area lab, with Claude handling data searches, hypotheses and analysisAnthropic's Claude AI model has identified a previously uncharacterised enzyme system in bacteriophages that shares some features with CRISPR, after AI agents searched more than 200,000 reverse transcriptases, Interesting Engineering reported. The finding offers an early example of AI agents contributing directly to biological discovery, although researchers stress the system's function is not yet known.The work formed part of Anthropic's new life sciences research program. Around 950 Claude agents spent 21 hours searching genomic data, using roughly 210 million tokens. They gathered more than 200,000 reverse transcriptases, flagged about 3,500 candidate systems and narrowed the field to 20 for detailed reports. Anthropic said genome mining on this scale could take an expert scientist weeks or months.The standout candidate, named array-associated reverse transcriptase, or ART, pairs a reverse transcriptase with a neighbouring partner gene and a long array of evenly spaced DNA repeats. One agent spotted the repeat array while examining raw DNA around an unusual enzyme family, then counted the repeats, measured their spacing, compared the arrangement with known systems and checked the scientific literature before flagging it for human review. The enzyme itself had been identified before, but the surrounding system, including the repeat array and an accessory protein of unknown function, had not.Early laboratory work found that the ART repeat array is expressed as multiple distinct short RNAs. That is significant because CRISPR arrays also produce RNAs that make CRISPR-Cas systems programmable. Anthropic said the combination of features resembles a small group of known systems capable of programmable operations on DNA, such as cutting, copying or inserting genetic material.Researchers caution that ART has not been shown to work like CRISPR, and experiments are under way to determine what its components actually do. MIT and Broad Institute professor Feng Zhang, a pioneer of CRISPR gene editing, reviewed the work and described it as an exciting example of AI agents aiding biological discovery that merits further investigation.Anthropic has also set up a molecular biology laboratory in the Bay Area to test candidates generated by computational searches, with human scientists carrying out all experiments and Claude used for searching data, generating hypotheses and interpreting results. Whether ART proves to be a new biotechnology tool or a biological curiosity, the approach points to AI agents compressing the early discovery phase of life sciences research. This article was written by Eamonn Sheridan at investinglive.com.