Digital Science connects AI agents to research data via Dimensions MCP

Digital Science has launched two new Model Context Protocol (MCP) servers for its Dimensions research platform, giving enterprise artificial intelligence agents direct, license-aligned access to a vast interconnected database of scientific and technological information. The integrations are designed to allow AI systems to retrieve and analyze current research data rather than relying solely on general-purpose models whose training data may be outdated, incomplete, or difficult to verify.

The release introduces the Dimensions Semantic Search MCP and the Dimensions Analytics MCP, two complementary systems aimed at different stages of research intelligence. The semantic search service is built for precision retrieval across more than 40 life science domains, while the analytics service connects AI agents to more than 430 million linked records covering publications, grants, patents, clinical trials, datasets, and policy documents. Together, the systems are intended to combine detailed evidence retrieval with large-scale mapping of the research landscape.

The technology is based on the Model Context Protocol, an open standard that enables AI applications to connect with external tools and structured information sources. Instead of asking an AI model to generate an answer from its internal parameters alone, an MCP-enabled agent can send a request to an authorized data service, receive relevant records, and use those records to construct a response. This approach can improve traceability and reduce the risk that an AI system will produce plausible but unsupported claims.

For scientists and research-intensive organizations, the distinction is significant. Large language models can summarize concepts and generate hypotheses, but they may not know about recently published findings, newly awarded grants, emerging patent activity, or developments in clinical research. Connecting agents to live databases allows them to retrieve information at the moment it is needed. Existing Dimensions API customers can use the new MCP integrations without obtaining an additional license, according to Digital Science.

The Dimensions Semantic Search MCP is designed to search scientific information by meaning rather than by exact word matching. Conventional keyword systems may miss relevant documents when researchers use different terminology, abbreviations, chemical names, or disease classifications. A semantic search engine analyzes the conceptual relationships within a query and uses domain ontologies—structured representations of scientific terms and their connections—to identify related evidence across publications, patents, grants, and clinical trials.

Digital Science gives the example of a search for “PFAS,” a broad class of per- and polyfluoroalkyl substances. A keyword search may prioritize documents containing the exact abbreviation, whereas an ontology-supported semantic system can identify related substances such as PFOS, PFOA, and PFHxS. The underlying process is not simply a similarity calculation between sentences. It involves mapping terms to concepts, recognizing hierarchical and associative relationships, and retrieving passages in which those concepts are discussed, potentially at the section level of individual documents.

This capability could be particularly useful in drug discovery, pharmacovigilance, medical affairs, biotechnology, and regulatory intelligence. An AI assistant could search for evidence involving a disease subtype, compound family, or drug class, then identify recurring relationships across millions of documents. Such searches may help researchers review safety signals, trace drug–disease associations, monitor therapeutic pipelines, and locate evidence that would be difficult to uncover through manually assembled keyword lists.

The Dimensions Analytics MCP serves a broader function by linking research outputs with the people, organizations, funding sources, and intellectual property associated with them. An enterprise agent could use the service to compare competitors in a therapeutic area, identify leading investigators, examine the activity of particular funders, or track research trends across regions and technology sectors. Because the database connects different record types, a query can move from a publication to its authors, institutions, grants, patents, and related clinical trials without requiring researchers to consult multiple disconnected systems.

Digital Science says the integrations are compatible with major AI platforms, including Claude, ChatGPT, and Gemini, and are intended to operate within existing enterprise workflows without custom integration work. For research and development teams, the practical appeal is the ability to ask an AI agent to perform tasks such as competitive landscape analysis, funding intelligence, technology-transfer scouting, or horizon scanning while drawing on structured records that can be inspected and verified.

Sebastian Schmidt, executive vice president of Enterprise at Digital Science, described the MCP servers as a bridge between artificial intelligence and authoritative research intelligence. Peter Haase, vice president of Knowledge Graph Technologies, emphasized that semantic retrieval allows systems to search scientific concepts rather than depending on users to predict every relevant term. The launch reflects a broader shift in scientific computing: AI agents are moving from isolated language interfaces toward connected systems that can retrieve, interpret, and analyze live evidence. If these connections are implemented with appropriate licensing, provenance controls, and human oversight, they could make complex research intelligence faster to obtain while preserving a clearer link between an AI-generated conclusion and the data supporting it.

Web References: https://www.dimensions.ai/products/all-products/dimensions-mcps/ ; https://www.dimensions.ai/

Keywords

Artificial intelligence, Model Context Protocol, MCP servers, Dimensions, semantic search, research databases, scientific data, data analysis, drug discovery, pharmaceutical industry, biotechnology, research analytics, clinical trials, patents, grants, scientific publishing