Intelligent automation startup Celigo Inc. has announced the launch of its Model Context Protocol Management platform. This innovative solution will enable information technology and security teams to govern the actions of artificial intelligence assistants and agents within enterprise systems.
The Model Context Protocol (MCP) has emerged as a standard for connecting AI assistants and agents to various business systems and services. Many enterprise services currently host MCP servers that facilitate access to these systems, providing tools for agents to interact with both basic and complex controls.
In 2024, Anthropic PBC introduced MCP as an open-source standard, often likened to a USB-C port for AI. This protocol serves as a universal access point, allowing AI models and agents to connect to databases, local files, search engines, business data, and application programming interfaces (APIs) to access essential resources.
Use cases for MCP include retrieving customer information from customer management systems, checking invoices within enterprise resource planning systems, updating order statuses in parcel tracking systems, and managing tickets in service desk applications. Tasks traditionally managed by integration specialists can now be executed autonomously by AI agents like Claude, ChatGPT, Copilot, and Cursor.
However, this capability generates significant security risks and access concerns, particularly if these agents operate outside their intended roles or possess unauthorized access. Therefore, it is crucial to define the limits of their activities and provide guidance on their operational scope. Although AI agents can act independently and wield considerable power, they are also prone to errors—commonly referred to as hallucinations—when functioning with minimal oversight, and they may be susceptible to exploitation by malicious actors.
“AI assistants and agents represent a new class of participants in the enterprise, and traditional models of shared credentials and overarching access are insufficient for their governance,” stated Chief Product Officer Matt Graney. “Enterprises must monitor who is behind each AI request, the permissions granted to them, and the actions they undertake.”
Celigo’s MCP Management offers IT teams a streamlined and governed pathway to regulate agents, ensuring they remain confined to their designated domains. Graney emphasized that each request is linked to a specific individual, with access authorized one tool at a time, allowing teams to trace AI interactions with different systems and maintain full transparency.
This centralized control framework enables comprehensive and granular management of AI assistants and agents across the enterprise landscape.
The company noted that this management interface connects MCP servers to corporate knowledge and integrates with over 1,000 applications tailored for business operations. It provides teams with the tools to manage user access to specific applications via an MCP Gateway and discover each MCP server through a centralized registry.
“Celigo’s MCP was the missing piece,” remarked Antoine Millet, Head of IT at Nidecker SA. “It serves as the intermediary connecting all our third-party tools in the cloud. We can now oversee the start, middle, and end processes within a single dashboard.”