---
title: "Fastal MCP Platform"
description: "Framework that exposes chosen functionality of any management system, legacy or modern, to AI agents through the standard Model Context Protocol."
url: "https://www.fastal.it/en/products/mcp-platform/"
lang: "en"
---

## A bridge between management systems and artificial intelligence

Fastal MCP Platform is a proprietary framework that enables creating an interface layer between your management systems and the world of AI agents. Through the standard MCP (Model Context Protocol), you can expose selected and specific functionality of any system, legacy or modern, in a controlled and secure manner.

### The problem

Companies have established management systems containing critical data and processes. Conversational artificial intelligence offers new interaction modalities, but connecting these worlds presents significant challenges:

- Legacy systems are not designed for conversational interfaces
- Exposing all management system features is risky and inefficient
- Each AI agent has different integration requirements
- Security and access control are fundamental

### The solution

Fastal MCP Platform creates an abstraction layer that:

- **Selects** — Expose only the features you want to make accessible, not the entire system
- **Standardizes** — Transform heterogeneous operations into uniform and documented MCP tools
- **Protects** — Maintain control over who accesses what, with complete logging
- **Enables** — Allow natural language interaction with your systems

### Two use case scenarios

#### Agentic interfaces for management systems

Develop conversational assistants that allow users to interact with the management system using natural language. Operators can query data, execute operations, generate reports simply by "talking" to the system.

#### Connectors for consumer agents

Create custom MCP connectors that allow consumer agents like Claude Desktop or ChatGPT to access your management system features. Users can thus leverage the power of these agents while maintaining the context of business data.

### How it works

1. **Analysis** — We identify the management system features to expose
2. **Mapping** — We define how to translate operations and data into MCP tools
3. **Implementation** — We create the interface layer with the framework
4. **Configuration** — We set up permissions, filters and access rules
5. **Deployment** — The MCP server is ready to connect to agents

### Supported systems

The framework is designed to integrate with any system:

- ERP management systems (SAP, Oracle, Microsoft Dynamics)
- Legacy systems with proprietary interfaces
- Applications with REST or SOAP APIs
- Relational and document databases
- Custom internally developed systems

### Why MCP

The Model Context Protocol is the emerging standard for communication between LLMs and external systems. Adopting it means:

- Compatibility with a growing ecosystem of AI agents
- Interoperability between different providers (Anthropic, OpenAI, others)
- Future-proof architecture that evolves with the standard
- Active community and complete documentation

## FAQ

### What problem does the platform address?

That established management systems hold critical data and processes but were never designed for conversational interfaces, that exposing all of a system's functionality is risky and inefficient, and that every AI agent has its own integration requirements. The framework builds the missing layer between the two worlds.

### Does the management system have to be rewritten to use it?

No. The framework is designed to integrate with any system without touching existing code: ERPs such as SAP, Oracle and Microsoft Dynamics, legacy systems with proprietary interfaces, applications with REST or SOAP APIs, relational and document databases, and custom in-house systems.

### Which parts of the management system get exposed?

Only the ones chosen. The layer selects a subset of operations, standardizes them into uniform, documented MCP tools, and keeps control over who accesses what, with full logging of permissions, filters and rules.

### What are the intended use cases?

Two. Internal conversational assistants, letting operators query data, run operations and generate reports by talking to the system. And custom MCP connectors that open the management system's functionality to consumer agents such as Claude Desktop or ChatGPT.

### How do you get to a running MCP server?

In five steps: analysis of the functionality to expose, mapping between the system's operations and MCP tools, implementation of the interface layer with the framework, configuration of permissions, filters and access rules, and finally deployment of the server, ready to connect to agents.