Skip to content

Your progress

Foundational0%

Deep0%

Progress tracking is for signed-in learners. Sign in (free) to track your Foundational and Deep progress on any device.

1Foundations

2Machine learning fundamentals

3Neural networks and embeddings

4Attention, transformers and LLMs

5LLM engineering

6AI applications

7Retrieval-augmented generation

8Evaluation

9Agents and MCP

Set up for Unit 9: MCP and agent tooling
Agents from first principles
Tool design for agentsSoon
Multi-step agents: planning, state and recoverySoon
SAP tools for agentsSoon
The Model Context Protocol (MCP)Soon
Agent frameworks comparedSoon
Joule and Joule StudioSoon
Joule agents and agent orchestration in SAPSoon
One order-exception agent, built three waysSoon

10Production and LLMOps

Set up for Unit 10: observability and containersSoon
Observability for AI systemsSoon
Latency budgets and performanceSoon
Token economicsSoon
Caching for LLM applicationsSoon
Model selection and routingSoon
Running AI on SAP BTP in productionSoon
The AI business caseSoon

11AI security

Set up for Unit 11: security testing toolsSoon
Prompt injection and tool poisoningSoon
Guardrails and content filteringSoon
Data security and PII in AI systemsSoon
Agent permissions and SAP authorizationsSoon
Red-teaming your AI systemSoon

12Models and advanced AI

Set up for Unit 12: local modelsSoon
When to fine-tune (and when not to)Soon
Quantization and local modelsSoon
Document AI, built and boughtSoon
Tabular AI and SAP's predictive modelsSoon
Embedded AI across SAP applicationsSoon
Buy vs. build for SAP AISoon

13AI-assisted SAP engineering

Set up for Unit 13: ABAP development toolsSoon
Coding agents for SAP developersSoon
ABAP Cloud and RAP for AI buildersSoon
An agent that calls a RAP serviceSoon

14The FDE operating system and capstone

Set up for Unit 14: capstone environmentSoon
Discovery: interviewing the customerSoon
Problem framing: from 'we need AI' to a measurable problemSoon
AI opportunity assessment: AI, automation or neither?Soon
Rapid prototyping: one day, one week, two weeksSoon
Technical storytelling: one architecture, five audiencesSoon
Adoption and change management for AISoon
Capstone: an enterprise exception management platformSoon

15Portfolio projects

Project: SAP master data semantic searchSoon
Project: late delivery predictorSoon
Project: SAP procurement copilotSoon
Project: finance variance assistantSoon
Project: invoice exception copilot (three-way match)Soon
Project: SAP configuration assistantSoon

•What's changing

New topics land here as SAP and AI change.

About Orchestrate

Orchestrate teaches enterprise AI through the lens of SAP: how the underlying AI works, how to build it yourself, and how SAP implements the same idea. The goal is people who can walk into an SAP customer, find where AI moves a business number, and ship it: forward deployed engineers.

Two layers, one topic

Foundational covers what the idea is, why it matters to the business, how SAP offers it, and the questions to ask. Deep covers how it works, how to build it, how SAP implements it, and what goes wrong.

How topics are made

A new topic is researched and drafted every weekday with AI assistance, reviewed by a human before it goes live, and cites its sources. The curriculum is a fixed spine of fourteen units, six end-to-end portfolio projects, and a What's changing track that is re-evaluated every week as SAP and the AI field move.

Feedback and corrections

Found something wrong or out of date? Please report it; corrections are prioritized over new topics.

Independent learning resource. Not affiliated with, sponsored by or endorsed by SAP SE. SAP and other SAP products mentioned are trademarks of SAP SE in Germany and other countries.