LLM Tool Use / Function Calling in Agents
Tutorials that teach this
- Tutorial Use Joule Studio, Classic Edition to Create an Agent
- Tutorial Optimize Prompts for Tool Calling Using a Custom LLM-as-a-Judge Metric in SAP AI Core
- Tutorial Add MCP capabilities to a CAP service
- Tutorial 6 - Create Agent to Coordinate Skills
- Tutorial Prompt Optimization with Tool Calling
- Tutorial Prompt Optimization with Tool Calling and Response Formatting
Code samples embodying this
Prerequisites
- Concept Model Context Protocol (MCP) Server Integration **Model Context Protocol (MCP)** is an [open standard that defines how AI models and agents can discover, understand, and interact](https://architecture.learning.sap.com/docs/ref-arch/RA0029/10-third-party-mcp-access/readme) with tools, data sources, and services. Developers use it to expose capabilities — such as SAP BTP account administration or the hana-ml Python library — as MCP endpoints that AI agents can consume in a standardized way. For example, the [HANA AI Toolkit local MCP server](https://architecture.learning.sap.com/docs/ref-arch/RA0033/readme) turns the hana-ml library into an MCP-accessible endpoint, enabling third-party agents and AI models to invoke its functionality without custom integrations.
- Concept Large Language Models (LLMs) The concept provided is **undefined**, and the supplied sources do not describe a specific, identifiable SAP developer concept by that name. Based on the available snippets — which cover topics such as [Retrieval Augmented Generation (RAG)](https://architecture.learning.sap.com/docs/ref-arch/RA0005/3-retrieval-augmented-generation/readme), prompting LLMs in the generative AI hub, and using custom or small language models on SAP AI Core — no definition can be responsibly constructed for an unnamed or undefined concept. Please provide a valid concept name so that an accurate, source-grounded definition can be written.