Model Context Protocol (MCP)

The Model Context Protocol (MCP) is an open standard that allows AI assistants and language models to access external data sources and tools. Anthropic released MCP in November 2024 and made it available as an open-source project. The standard regulates how a language model sends a request to a database, a file system or an enterprise system and receives the response back in a structured format. This means a language model no longer needs to be specially customised for every single data source.

The architecture follows a client-server model. An MCP server provides data or functions, such as access to a CRM, a calendar or a database. An MCP client, embedded in an AI application, connects to this server and retrieves the provided functions. Before MCP, a separate connection had to be programmed for every combination of AI application and data source. MCP replaces this multitude of individual solutions with a common protocol, much like a USB socket serves different devices via the same connection.

MCP in the SAP Business One context

For SAP Business One, MCP is relevant as soon as an AI assistant needs to access live business data, such as open items, item stock levels or document data via the Service Layer or the DI-API. Instead of building a separate interface for each type of request, an MCP server provides the relevant B1 functions once. Any MCP-enabled client, whether a chat interface or developer tool, can access it.

Demarcation

MCP is not a replacement for the service layer or a REST API. It sits rather as a layer above that: MCP standardises how an AI model calls such interfaces, not how the interface itself is built. Compared to retrieval-augmented generation, MCP covers a different need. RAG retrieves text knowledge from a knowledge base, whereas MCP gives the model access to structured data and executable functions.

AI in everyday work: Person starts with their own task instead of the learning pile

AI in everyday work: Here's how to find the right start

Anyone who wants to use AI in their daily work is familiar with this: You read an article about prompts. A new model appears in it ...

AI assistant meets SAP Business One: Two paths out of the Versino AI shop

AI assistants like Claude or Copilot write emails, compile documents, and explain technical terms. But when you ask them about the open-source code...

Sprint and Swirl: How an AI agent divides work and learns from trials

This series deals with artificial intelligence in conjunction with SAP Business One. It takes a close look at fundamental AI principles...

LATS: Considering several paths simultaneously with AI — and the availability check in SAP B1

This series focuses on artificial intelligence in conjunction with SAP Business One. Not as a collection of product announcements, but as...

ReAct and RLEF: How Artificial Intelligence Learns from Real Mistakes — and What Automated Bank Reconciliation in SAP Business One Has to Do with It

This is the fourth instalment in a series on this blog dealing with AI fundamentals in conjunction with SAP Business One...

SAP B1 10.0 FP2608: Service Layer AI as a transactional layer

The SAP Business One Service Layer has previously served predominantly as a passive data provider: applications requested data via OData, each ...