Anthropic official brand assets (anthropic.com)
Anthropic's biology lab used its Claude model to search the literature, refine lab methods, and map a mysterious protein barrel from a bacteriophage
Anthropic’s biology research lab says it has produced the first 3D structure of a vault particle found in prokaryotic cells. Its AI model, Claude, helped at nearly every stage of the work.
Vaults are giant, hollow protein barrels that show up in cells across the tree of life. Scientists have known about them since 1986. They still don’t know what vaults actually do.
What Anthropic actually did #
The specific vault in question comes from a bacteriophage. That’s a virus that infects bacteria, a category of life that sits far from the human cells where vaults are better known.
Until now, structural work on vaults has focused on eukaryotic cells. Those are the complex cells that make up plants, animals, and fungi. Anthropic’s result is the first 3D structure drawn from the prokaryotic side of the ledger.
Anthropic announced the finding on November 4, 2026. The company described Claude’s role across three distinct steps of the project.
First, the literature search. Claude combed through sequence literature to help identify where vault-like proteins might exist.
Second, the lab method. The model helped optimize a technique for isolating the particles.
Third, the structure itself. Claude assisted with mapping the vault’s 3D shape.
AI, tech, and the markets they move—in one daily briefing.
Daily. Free. Join 34,000+ readers across crypto, finance, and policy.
Days instead of months, with a human in the loop #
The headline efficiency claim is about time. According to the research findings, tasks that would usually take weeks or months, including literature reviews and structural mapping, shrank to days.
Anthropic’s lab stressed that human oversight remained essential. The model made occasional errors, and researchers had to catch and correct them.
Why vaults have been such a puzzle #
Vaults have been sitting in plain sight for decades. They are large, distinctive, and widespread. Yet their biological role has stayed largely unspecified since their 1986 discovery.
The new structure doesn’t solve that mystery outright. But the research findings suggest it may offer hints, pointing toward possible links to cellular transport and stress responses.
The research findings note this work could eventually reshape understanding of molecular biology in both prokaryotic and eukaryotic systems.
Anthropic’s growing bet on biology #
Anthropic described the vault result as “another finding” from its biology research lab, signaling an ongoing program rather than a single experiment.
The company established a wet lab in 2026. A wet lab is a physical space where researchers handle real biological samples, as opposed to running purely computational models.
What this means for AI-assisted science #
For the scientific community, the most practical takeaway is the workflow. Anthropic laid out a template where an AI model touches the literature review, the lab protocol, and the structural analysis in a single project. Anthropic itself flagged that Claude made mistakes that humans had to fix. In biology, an uncaught error in a method or a structure can send a research program down the wrong path for a long time.
For now, a protein barrel that has stumped researchers since 1986 has a new 3D portrait. Its purpose remains an open question, but scientists now have a much clearer picture of what they’re trying to explain. Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.