Watch on YouTube
Watch on Vimeo
In this session, learn how the Model Context Protocol (MCP) tackles the challenges of utilizing unstructured data by providing seamless, permission-aware integration for AI models and data sources, eliminating the need for intricate custom connectors. Discover how this ‘USB for AI’ enables enterprise-wide data interaction and management, offering a reliable and future-proof architecture.
CTERA addresses the problem of connecting enterprise data sources to Generative AI models, which traditionally required custom connectors for each application, resulting in exponential complexity and fragility. The MCP protocol offers a solution by providing a seamless, guaranteed integration between any Gen AI model and tool that supports MCP, while also being permission and identity aware. It ensures contextual information about the user, their permissions, and authentication is readily available. CTERA has embraced MCP as a core part of its strategy, implementing it across its products.
CTERA’s implementation of MCP is structured in two main layers. The first layer resides within the global file system product, where files are stored, enabling Gen AI agents to access and utilize data while respecting user permissions. The second layer focuses on data intelligence, providing a semantic layer over the data that exposes textual information and metadata through MCP. The MCP server is implemented within the enterprise application, while the MCP client is the AI tool. This architecture is not specific to any LLM and supports OAuth2 authentication, allowing for secure and permissioned access to data.
A demonstration highlighted how CTERA’s MCP server could be easily enabled via the user interface, showcasing its integration with Claude. The demonstration showed how a user could instruct Claude to interact with the global file system, list files, read them, summarize them, and write the summary back, all without writing any code. This example illustrated how MCP enables end-to-end applications that democratize access to data and allow users to simplify repetitive tasks, thereby increasing efficiency and job satisfaction.
Personnel: Aron Brand
Thank you for being part of the Tech Field Day community! Our mailing list is a great way to stay up to date on our events and technical content, and we appreciate your signup.
We promise that we’ll never spam you, send ads, or sell your information. This list will only be used to communicate with our community about our events and content. And we’ll limit it to no more than one message per week.
Although we only need your email address, it would be nice if you provided a little more information to help us get to know you better!
