REAV

Glossary

MCP

Model Context Protocol

An open protocol that lets an AI assistant connect to external tools and data sources through one common interface instead of a bespoke integration each time.

MCP was published by Anthropic in November 2024 and has since been adopted well beyond it. The problem it addresses is combinatorial: each pairing of an assistant with a system it might read from is a separate integration. MCP defines a common interface so a system exposes its capabilities once, and any client that speaks the protocol can use them.

A server exposes three kinds of thing: resources the client can read, tools it can call, and prompts it can reuse. The client is the assistant; the server sits with the system being exposed.

The part that decides whether it is safe to use is permissions, and they belong on the server side. What the assistant can see and what it can change are set by the system exposing them, not by the model - which is the only arrangement where the answer to what can this thing touch does not depend on the model behaving well.

For an operation running an ERP, the practical consequence is that asking questions of production data stops requiring a report to be built first. The corresponding discipline is that read access and write access are separate decisions, and that write actions are worth routing through a human.

How the REAV ERP exposes its data over MCP

Related terms

  • RAG Answering a question by first retrieving relevant passages from your own documents, then generating the answer from those passages rather than from the model memory.
  • SOP A written, versioned instruction that defines how one operation is performed, so any trained technician produces the same result.
  • Audit trail A chronological record of who did what, when, and to which record - kept so that it can be checked later and not quietly altered.

← All glossary terms