MCP agents in the enterprise: how SimpleOne and Ainergy built AI into business processes

Enterprise AI agents can already write emails, summarize meetings and find answers in a knowledge base. But the SimpleOne team, together with its technology partner Ainergy, has gone further. They built MCP agents into the enterprise platform itself, so that AI does not merely offer advice alongside processes but becomes a full participant in them.

The limits of the text-based agent

A standard AI agent in a corporate environment is a smart chatbot sitting next to the system, not inside it. It helps phrase a task, summarize correspondence or find the right section of documentation. But creating a ticket in ITSM, checking the status of an incident, finding a related object in a database or querying an external HR system is beyond its capabilities. As a result, employees still switch between the agent and the corporate system, copying data back and forth by hand. AI helps people think, but not do.

For enterprise scenarios such as ITSM, SDLC or asset management, this is a fundamental limitation: what matters there is not text but a concrete action in the system. This contradiction, the agent as consultant versus the agent as executor, became the starting point for SimpleOne and Ainergy.

MCP: managed access to actions

Model Context Protocol (MCP) is an open standard developed by Anthropic that formalizes how a language model interacts with external tools and systems. The key difference from direct API calls is that the agent does not access the database directly. Instead, it receives a set of formally described tools with clear access rights and explicit execution logic.

Each tool is described as a contract: what it accepts as input, what it returns, and what permissions are needed to call it. The LLM itself chooses the right tool, calls it and passes the result to the user, all within a managed perimeter and without direct access to the system's internal architecture.

In an enterprise context, the set of MCP tools for an ITSM platform can include:

• retrieving data on a ticket, incident or system object
• checking the current status of a task or change
• creating a new record, such as a ticket, a task or a related object
• searching for related objects by given criteria
• querying an external system (ERP, HR, CMDB) and returning the result to the agent

How it works in SimpleOne

In the SimpleOne platform, the MCP server acts as a layer between the language model and the business logic. Ainergy developed the infrastructure layer of the GenAI platform, which makes it possible to describe MCP tools at the level of the SimpleOne object model, without giving the LLM direct access to the database or the platform's internal logic.

In practice, this means the same agent is configured differently for different roles. An IT specialist sees tools for working with incidents and changes, a project manager sees tools for tasks and statuses, and an external user gets only a limited set for submitting requests. Permissions to call tools are configured separately and tied to the user's role or the context of the operation. The agent sees only permitted actions and works strictly within the limits set by the security team.

For an enterprise environment this is essential: auditability, security and manageability are preserved, while the AI gets real tools to take part in processes, not as a consultant but as an executor with strictly limited rights.

What this means

MCP architecture changes the role of AI in the enterprise, from a text assistant to a participant in workflows. For ITSM, SDLC and other enterprise-class platforms, this opens a path to automating routine operations without compromising control over access rights and data security.

Companies get AI that does not just think alongside a process but acts inside it, within managed and auditable boundaries.