In exclusive interviews announcing the new partnership between the Allen Institute and the Howard Hughes Medical Institute, Anthropic’s head of life sciences Jonah Cool and Grace Huynh, executive director of AI applications at the Allen Institute, said elite science labs are using Claude-powered AI agents to solve analysis, annotation, and coordination bottlenecks to years of research timelines.

A ‘compressed 21st century’

Cool, a cell biologist and geneticist by training as well as a technology leader speaks luck that he was inspired by a 2024 essay by Anthropic CEO Dario Amodei, Machines of Loving Gracearguing that “AI-enabled biology and medicine will allow us to compress the progress that human biologists would have achieved in the next 50 to 100 years into five to 10 years.”

It’s an idea that Amodei describes as a “compressed 21st century” that could make possible everything from near-universal prevention of infectious disease and major reductions in cancer deaths to effective treatments for genetic disorders, Alzheimer’s, and other chronic diseases. Amodei also suggested that AI could enable more personalized therapies, expand human control over biology itself, and even dramatically extend healthy lives.

For Cool, that vision directly maps to the use of AI agents in science — not as tools that deliver breakthroughs, but as systems that can take the time-consuming analysis, coordination, and experimentation tasks that slow discovery across labs, allowing humans to potentially make critical new discoveries.

“What AlphaFold has achieved is incredible,” said Cool, referring to the system’s solution to the long-standing problem of protein folding. “But we’re talking about something different here. It’s about working with process science teams and embedding AI in their daily work.”

Huynh said the move toward AI agents at the Allen Institute, a non-profit bioscience research organization founded in 2003 by Microsoft cofounder Paul Allen, built tools that are already used by many researchers, especially Anthropic’s Claude Code, which has become popular among computational biologists. In addition, the goal, he said, is not to use AI everywhere, but to focus on specific parts of the research process — such as data analysis tasks that can take months — where the agents will have the most practical impact and significantly accelerate scientific work.

No single researcher can find every connection

We’re starting to get to a point where ‘big science’ is the norm,” he said. Scientists are generating so much data today—from single-cell genomics and massive imaging datasets to connectomics, the study of how neurons are connected in the brain and nervous system—that no single researcher can hold it all in their head or see every connection.

Cool pointed to the Allen Institute and the Howard Hughes Medical Institute as ideal partners because of the role they played in shaping modern science. The Allen Institute produces some of the most widely used biological datasets in the world, including detailed maps of the mouse brain that show where genes are active in actual tissue—resources that have become standard tools for researchers in all fields, not just neuroscience. Recently, such maps have been pushed to single-cell resolution, which dramatically increases their scientific value while also making them more complex to analyze.

And at HHMI’s Janelia Research Campus, researchers have developed foundational tools such as calcium indicators such as GCaMP, which allow scientists to view neurons in real time, and advances in super-resolution microscopy that have helped push the physical limits of light imaging. The emphasis on tools and data, Cool says, is exactly what makes these institutions fertile ground for AI agents: facilitating analysis, annotation, and coordination there doesn’t just help a lab—it spills over into the whole of science.

“Science is an interesting but very repetitive and often very tiring practice,” he explained. “Increasingly in science, what that means is a lot of work related to analyzing and transforming data sets,” he said. “I think we’re kind of approaching a world where it’s going to take a lot more work, but…you get to the next steps and the experiments are much, much faster.”

A future where AI can help make predictions

Cool also describes a future where AI agents will not only analyze the results, but help scientists decide which hypotheses to pursue—narrowing hundreds of possible experiments down to a few of the most worthy runs, and even suggesting new DNA designs based on patterns that are not readily apparent to humans.

“We’re moving forward with models that help make predictions,” by enhancing the knowledge people already have, he said. “We started with, ‘help me prioritize the hypotheses I have,’ because I have a limited amount of resources, and I want to do all 100 experiments, but I only have money for 10.”