As AI capabilities evolve from simple question-and-answer tools to sophisticated agents that can interact with systems and data, we need new infrastructure to support these interactions – and that’s where MCP comes in.
To make informed decisions about your content, your data, and your distribution methods, it’s important to get acquainted with this latest development and how it impacts scholarly publishing.
APIs: The Foundation We Already Know
To understand MCP, we can begin with something more familiar: APIs. An Application Programming Interface (API) is a set of rules that allows different software systems to communicate with each other. Think of APIs as standardized connectors that let applications exchange information without needing to understand each other's internal workings.Publishers already rely on APIs extensively, as they enable the interoperable, modular systems that power scholarly publishing today. When a library's discovery system pulls article metadata from your platform, that's an API. When you integrate DOI registration with Crossref, that's an API. When analytics tools access usage data, they're using APIs.
MCP: APIs Designed for AI Agents
Model Context Protocol serves a similar connective function, but it's specifically designed for AI systems. While traditional APIs were built for application-to-application communication, MCP creates standardized pathways for AI models to both access external data sources and take actions through external tools.MCP reduces the need for custom integration code (as with APIs) by providing a universal standard. AI systems can discover available data sources and tools, understand their capabilities, and interact with them through one consistent protocol.
What MCP Actually Does
MCP operates through a client-server architecture with three core capabilities:- Tools are model-controlled—the AI model itself decides when to invoke them based on analyzing user requests. When a researcher asks "What are the citations for this paper?", the AI determines that a citation retrieval tool is needed.
- Resources are application-controlled—the host application (like Claude Desktop or an IDE) or the user explicitly adds them as context, similar to attaching files.
- Prompts are user-controlled templates that users select through commands like "/generate-literature-review."
Practical Applications for Publishers
For publishers, MCP opens intriguing possibilities across multiple dimensions. For content discovery, researchers and non-specialists alike could describe their information needs using natural language queries, with AI systems using MCP to search across subscribed content while applying institutional access rules seamlessly.With authentication covered, MCP has the potential to enhance existing business models or even create new ones. Organizations licensing content through corporate subscriptions could integrate your material directly into their AI-powered research assistants, creating new channels for content delivery and making enterprise AI integrations more practical to pursue at scale.
The Infrastructure Questions
Like any emerging technology, MCP raises important questions for publishers. How do we maintain content security while enabling AI access? What metadata standards will AI systems need to effectively use our content through MCP? How do we ensure that AI-mediated discovery still generates appropriate usage tracking and attribution? What does licensing look like when content flows through AI agents rather than traditional interfaces?These aren't just technical questions: they're strategic ones that will shape how AI integrates into the scholarly communication ecosystem. At Silverchair, we're exploring these questions through our AI Lab, working with publishers to understand where AI-enhanced discovery can create genuine value and where it introduces unnecessary complexity and releasing new offerings after thorough testing and validation.
Looking Ahead
Model Context Protocol represents a shift in how we think about system integration. Rather than building point-to-point connections between specific applications, the future offers a more flexible infrastructure where AI systems can access information and perform actions dynamically based on user needs and permissions.For publishers, this suggests a future where content doesn't just sit in repositories waiting to be discovered. Instead, content actively participates in AI-assisted research workflows, delivered through authenticated, standards-based protocols that give publishers control over how their content is accessed and attributed. Whether MCP becomes the long-term standard or gives way to something else, the direction is clear: AI systems need structured, authenticated ways to access content. Understanding how these integrations work today prepares publishers for wherever the technology lands.
We have been actively exploring the best ways to leverage these technologies at Silverchair for 6 months, and we're committed to sharing what we learn. The future of scholarly publishing will be shaped not just by what AI can do, but by how we choose to connect it to the rich ecosystem of content and services publishers have built.
Want to explore how MCP and other AI technologies will work in your publishing environment? Reach out to the Silverchair AI Lab team to learn more about what we’re developing and practical ways you can explore with us.