Enterprise challenges in AI integration

Traditional AI workflows often struggle to access structured business data securely, creating challenges for organizations that want to use AI for operational tasks. While unstructured data such as documents, emails, and logs can be easily processed using embeddings and vector stores, structured business data remains a significant hurdle. Orders, inventory, customer records, and transactions stored in OpenEdge databases and governed by business logic are not readily accessible to AI workflows. Without any secure, standardized mechanism that allows AI agents to execute business operations and not just read data while maintaining compliance and governance, exposing sensitive business logic introduces risk and violates enterprise security policies.

Essential requirements for AI workflow integration

Organizations running OpenEdge applications require AI workflows that can:
  • Make intelligent decisions using proprietary business data and rules.
  • Execute business operations through natural language interactions.
  • Maintain data sovereignty and security compliance in on-premises or private cloud environments.
  • Audit and govern AI interactions with business systems to ensure accountability.

The foundation - Model Context Protocol

The Model Context Protocol (MCP) defines a standardized interface that enables Large Language Models (LLMs) to interact with external enterprise systems in a secure and predictable manner. MCP provides the following capabilities:
  • Tool discovery—LLMs can identify available business operations through structured metadata exposed by the MCP Server.
  • Context awareness—Each tool includes detailed information such as parameters, schemas, and business entity types, allowing LLMs to make informed decisions before execution.
  • Secure execution—Operations are performed under strict authentication and authorization controls to maintain compliance and protect sensitive data.
  • Structured responses—Responses are returned in a consistent, machine-readable format, enabling LLMs to process results and drive subsequent actions effectively.

The solution - OpenEdge MCP Server

The OpenEdge MCP Server addresses the challenges in AI integration by transforming an OpenAPI specifications and optional prompt assets into AI-accessible tools. This process makes each API operation discoverable and usable by AI workflows, without requiring manual coding, enabling secure and efficient interaction with enterprise APIs.

The OpenEdge MCP Server acts as a bridge between AI agents and OpenEdge services, exposing APIs as secure, discoverable tools. The OpenEdge MCP Server connects to PAS for OpenEdge services through REST and WEB interfaces, enabling seamless communication between OpenEdge applications and MCP components. The following examples outline how AI-enabled OpenEdge workflows can execute business operations through natural language requests:
Business operation Natural language request example
Order analysis Show me orders from the last 30 days with delivery delays and suggest remediation actions.
Customer onboarding Create a new customer account and set up their initial order based on our standard package.
Inventory optimization Analyze inventory levels across warehouses and recommend restock priorities.
For detailed information about the architecture and core components of the OpenEdge MCP Server, see OpenEdge MCP Server architecture.