Summary
A University of Illinois Urbana-Champaign team describes iCloudBiofoundry, a platform that links AI-generated research workflows to distributed robotic biofoundries. A bioRxiv preprint reports demonstrations in enzyme retrieval, protein engineering and metabolic engineering.
Researchers at the University of Illinois Urbana-Champaign have described iCloudBiofoundry, a cloud-accessible platform that connects AI-based research planning with experiments run on robotic biofoundries. The bioRxiv preprint reports demonstrations in enzyme retrieval, protein engineering and metabolic engineering.
The system is designed to narrow a persistent gap in autonomous experimentation: computational tools can propose research plans, but carrying them out requires laboratory procedures and physical equipment. iCloudBiofoundry brings those stages together in a workflow intended to operate with minimal human intervention.
Two AI layers connect plans to experiments
The platform assigns different roles to two components. iBioGenie, a scientific multi-agent system, takes a research objective expressed in natural language and develops testable hypotheses and experimentally grounded protocols. According to the authors, it draws on literature, databases and computational tools in preparing those plans.
BiofoundryAI provides the link to laboratory work. It translates the proposed protocols into executable workflows for distributed robotic biofoundries. The platform then collects experimental data for use in iterative design, creating a loop in which results can inform subsequent steps.
A biofoundry is a laboratory setup that uses automation to carry out biological experiments. Here, the significant architectural idea is the connection between software-generated experimental plans and robotic execution—not simply using AI to suggest a biological design.
Demonstrations span three kinds of biological work
The authors report end-to-end demonstrations covering enzyme retrieval, protein engineering and metabolic engineering. Together, these examples show the platform being applied to tasks involving finding enzymes and engineering proteins or metabolism. The preprint presents them as representative case studies of the system’s capabilities.
Linking planning, execution and data collection could make it easier to run repeated experimental cycles across computational and physical systems. That integration is relevant to research workflows in which proposed designs need to be tested in the laboratory before the next design step can be chosen.
The evidence is a bioRxiv preprint describing three selected case studies. It supports a report of workflow integration in those demonstrations; it should not be read as evidence that arbitrary biological research can already be automated.