Once learners understand LLM fundamentals, this module introduces the SAP environment used to work with generative AI models and enterprise AI workflows.

Participants explore the architecture and capabilities of SAP's generative AI hub, including model access and management, prompt-related functionality, and the Orchestration Service.

The uploaded curriculum specifically identifies understanding the hub's architecture, accessing and managing multiple LLMs, using prompt engineering and model management capabilities, and leveraging orchestration for secure enterprise AI applications as key outcomes.

SAP's current introductory course also covers integration with SAP AI Core and SAP AI Launchpad, access to multiple models, Model Library, Chat, Prompt Editor, and the Orchestration Service.

Organizations may need different AI models for different requirements. One model might perform better for summarization, another for reasoning, while another may provide a better balance between performance and cost.

SAP's generative AI hub provides a managed environment for accessing and working with generative AI models within SAP's enterprise AI ecosystem.

This unit helps learners understand its architecture and core capabilities.

Participants explore how the hub can support model discovery, model access, prompt experimentation, prompt management, and AI application development while maintaining enterprise requirements around governance and security.

Current SAP Learning material identifies interfaces such as Model Library, Chat, Prompt Editor, and Orchestration Service as part of this environment. 

Enterprise generative AI applications often require more than sending a prompt directly to a model.

A workflow may need to prepare the prompt, select a model, retrieve business context, protect sensitive information, apply safety controls, and process the resulting response.

The Orchestration Service helps bring these activities together into structured AI workflows.

Learners explore how orchestration can help developers build more controlled and reusable generative AI solutions.

Current SAP documentation shows that orchestration workflows can be built, tested, saved as configurations, and executed through SAP AI Launchpad and SAP AI Core. 

This module moves from understanding the platform to actually building and improving generative AI solutions.

Learners develop prompts for business scenarios, integrate LLMs into applications using the SAP Cloud SDK for AI, apply more advanced prompt engineering approaches, and compare prompts across multiple models.

The source curriculum identifies these four areas as the core units of the module.

SAP's current course also covers prompt templates and versions, Prompt Registry, orchestration workflows, advanced techniques such as few-shot and meta prompting, multimodal inputs, automated evaluation, and comparison of models based on factors including accuracy and cost. 

This unit takes prompt engineering from a general concept into the SAP generative AI hub environment.

Learners explore how prompts can be developed, tested, refined, saved, and reused for business applications.

Participants also develop an understanding of prompt templates, where dynamic information can be inserted into a structured prompt without rewriting the complete instruction each time.

For enterprise development, reusable prompts help teams maintain consistency across applications and use cases.

Learners also become familiar with concepts such as prompt versioning and the Prompt Registry, helping them understand how prompts can be managed as reusable development assets rather than temporary text entered into a chat interface. 

Experimenting with an LLM through an interface is useful, but enterprise applications need programmatic access.

This unit introduces the SAP Cloud SDK for AI and helps learners understand how generative AI capabilities can be incorporated into application development.

Participants explore how developers can interact with models, send requests, receive generated responses, and integrate AI capabilities into broader application workflows.

The unit bridges the gap between prompt experimentation and application development, helping learners understand how LLM functionality becomes part of a real enterprise solution.

Initial prompts rarely represent the best possible version of an AI interaction.

This unit develops more advanced prompt engineering skills and helps learners systematically refine model responses.

SAP's current course specifically identifies techniques including few-shot prompting and meta prompting.

Learners explore how examples, instructions, context, formatting requirements, and prompt structure can influence the model's output.

The emphasis is on controlled experimentation: adjust the prompt, evaluate the result, identify weaknesses, and refine the instruction until the response better satisfies the business requirement.

Different LLMs can respond differently to exactly the same prompt.

This unit helps learners compare model behavior rather than assuming a single model is ideal for every use case.

Participants explore how prompts can be tested against multiple models and how results can be evaluated according to factors such as response quality, accuracy, consistency, latency, and cost.

SAP's current course specifically includes comparing models for accuracy and cost and automating prompt evaluation.

The objective is to help learners make more informed model-selection decisions based on the actual requirements of a business scenario.

The final module moves into one of the most important areas of enterprise generative AI: grounding AI responses in trusted organizational knowledge.

General-purpose LLMs are trained on broad datasets. They do not automatically know an organization's private documents, current policies, internal procedures, product documentation, or other proprietary information.

Document grounding helps address this limitation.

Learners explore how documents can be converted into searchable vector representations, how relevant information can be retrieved according to a user's request, and how that information can be supplied to the LLM as context.

This forms the foundation of Retrieval Augmented Generation (RAG).

The source curriculum specifically covers document grounding, vector knowledge bases, configuration of the Document Grounding module in the Orchestration Service, RAG, and reliable enterprise AI solutions. 

This unit introduces the end-to-end concept of grounding generative AI responses using enterprise documents.

Learners explore how organizational information can be prepared as a vector knowledge base and made available for semantic retrieval.

SAP's current Document Grounding implementation follows a RAG approach and uses vector representations to retrieve relevant document information that can be supplied as context to an LLM.

A simplified flow looks like:

Enterprise Documents → Content Processing → Embeddings → Vector Knowledge Base → Semantic Retrieval → Relevant Context → LLM → Grounded Response

SAP's current documentation explains that document pipelines can fetch information, divide content into chunks, generate semantic embeddings, and store those representations for later retrieval.

This allows the application to retrieve information relevant to a user's question rather than passing an entire document repository to the LLM.

The result can be AI responses that are more closely connected to trusted organizational knowledge.

Generative AI skills are developed most effectively through experimentation.

The uploaded curriculum includes a practice environment where learners can complete hands-on exercises involving prompt development, template management, and workflow orchestration within SAP AI Launchpad.

SAP's current learning material confirms that this practice environment provides access to a preconfigured SAP AI Launchpad system for completing course exercises.