Researchers have introduced Denario, an artificial intelligence assistant designed to support scientists throughout the research process. Developed by a team from the University of Cambridge, the Flatiron Institute, and the Autonomous University of Barcelona, Denario uses large language models to help with tasks such as identifying new research questions, analyzing data, and drafting scientific documents.
Unlike existing AI tools that typically focus on one aspect of research at a time, Denario is structured to handle multiple stages. It can synthesize academic papers, propose novel hypotheses, analyze datasets, and write manuscripts. According to its creators, this modular system allows users to select specific components for their needs or use the tool for an end-to-end workflow.
“Sometimes the most interesting thing is the idea, because maybe it’s a new idea that hasn’t been explored,” said Dr Francisco Villaescusa-Navarro from the Flatiron Institute and one of Denario’s primary developers. “Sometimes it’s a new method that’s never been applied to a certain dataset. There are many ways Denario can help expand the way we think and point us in new directions.”
The researchers emphasize that Denario is not intended to replace human scientists. The current version has limitations: only about 10% of its outputs provide valuable insights and there have been instances where it fabricated data. Human oversight remains necessary.
Denario was developed by Dr Boris Bolliet from Cambridge, Dr Pablo Villanueva Domingo from the Autonomous University of Barcelona, and Villaescusa-Navarro. The project involved experts from various fields including astrophysics, biology, chemistry, neuroscience, mathematics, machine learning, quantum physics, and philosophy.
Recent advances in large language models like ChatGPT have inspired this effort to see how such technologies might assist across all phases of scientific work. Denario operates through a set of specialized AI agents dedicated to different tasks—such as coding or summarizing results—that can be used together or separately.
“We designed Denario with a modular architecture so that users can choose which of its components best fit their research, whether that’s coding, exploring research ideas, summarising results or something else,” said Bolliet from Cambridge’s Cavendish Laboratory.
To use Denario fully, researchers upload a dataset along with instructions describing their objectives. The system then generates potential project ideas by refining approaches to the data and reviewing relevant literature for novelty and context. Subsequent modules suggest analysis methods and execute them using CMBAgent—a multi-agent backend—before interpreting results and producing written summaries.
“The agents all work together to make it possible,” Villanueva Domingo said. He noted that users can review each module's output or run individual agents as needed.
Testing has included hundreds of end-to-end runs across disciplines such as astrophysics and materials science. While most outputs were not deemed suitable after expert review, about 10% yielded promising questions or findings.
The interdisciplinary capacity of Denario may help uncover research questions outside a specialist's usual perspective. “Denario can pull ideas from other fields that maybe a scientist is less familiar with and would never have considered,” said Villanueva Domingo. “That interdisciplinary nature is very exciting.”
Bolliet added: “I hope that Denario will help accelerate science by providing researchers with tools that allow them to spend less time on menial tasks — like scrolling the arXiv, formatting images, summarising analysis — and more time on deep creative thinking.”
Future plans include improving efficiency and output quality by filtering low-value results automatically. Challenges remain around technical accuracy—for example in expressing uncertainty—and ethical issues such as avoiding AI-generated false information (“hallucinations”) or addressing copyright concerns.
The team stresses ongoing collaboration between academia and industry was essential for developing Denario. They invite open discussion about responsible use of AI assistants in science while working on safeguards against misuse.
Boris Bolliet is also noted as a Fellow at Trinity Hall within Cambridge University.