The persistent identifier for research projects

In the traditional scientific record, information is often siloed by output. A researcher has an ORCID, a paper has a DOI, and an institution has a ROR. However, the actual connective tissue—the research project itself—has historically lacked a persistent, machine-readable identifier. RAiD (Research Activity Identifier) is an ISO standard (ISO 23527) and infrastructure service designed to solve this problem by providing a stable handle for research activities. Developed and administered by the Australian Research Data Commons (ARDC), RAiD functions as a metadata envelope that tracks the 'who, what, where, and when' of a project over its entire lifecycle.

While the concept of a persistent identifier (PID) might sound like dry administrative plumbing, it is becoming a foundational component for the next generation of automated research. As AI agents increasingly participate in the scientific process—performing literature reviews, generating hypotheses, and executing code—the need for a structured way to log these activities is paramount. RAiD provides the bucket in which these agentic actions can be collected and verified.

From human-centric to agent-ready metadata

The RAiD service is implemented through reference software known as Raido, which is available as open-source code on GitHub. This software allows research organizations to establish their own RAiD services, minting identifiers that can be updated as a project evolves. Unlike a DOI, which is typically minted once a paper is finished, a RAiD is intended to be minted at the start of a project. It grows over time, accumulating links to data sets, software repositories, and eventually, publications.

This longitudinal tracking is what makes RAiD particularly relevant to the AI agent ecosystem. When an autonomous agent is deployed to conduct a systematic review or run a series of simulations, its work can be signed and associated with a specific RAiD. This creates a clear provenance trail that allows future researchers—human or otherwise—to understand the context in which a specific discovery was made. In an era where AI-generated content can muddy the scientific record, the ability to anchor discoveries to a verified project identifier is a significant step toward reproducibility.

Governance and global standards

RAiD is not a commercial product in the traditional sense; it is a service managed by the ARDC, a national research infrastructure facility in Australia. The project represents a coordinated effort to modernize research data management on a global scale. By achieving ISO status, RAiD has moved from a regional experiment to an international standard, ensuring that the identifiers minted today will remain resolvable for decades.

Competitively, RAiD does not seek to replace existing PIDs but rather to orchestrate them. It acts as an aggregator, pointing to the various DOIs and ORCIDs that constitute a project's footprint. For organizations building research-oriented AI agents, integrating with the RAiD API provides a standardized way to handle project metadata without inventing a proprietary logging system. As the volume of research activity continues to scale beyond human capacity, the infrastructure provided by the ARDC and the RAiD standard will be what allows us to keep track of the machine-led discoveries to come.