Analytics Legends The knowledge platform for SAP Analytics
Academy module

Capstone — An End-to-End SAP AI Analytics Solution

Capstone — An End-to-End SAP AI Analytics Solution — Analytics Legends section illustration for the SAP Analytics knowledge base (concepts, studies, Academy)

As of 2026-09-25

Edited by , sole editor and publication director — 27 years in the SAP ecosystem. How these pages are produced, and where automation is used.

The capstone of the SAP AI path: a five-layer, end-to-end architecture combining SAP Datasphere/BDC as a grounding source, HANA Cloud's Vector Engine, the generative AI hub's fixed orchestration pipeline, a tabular prediction (SAP-RPT or PAL) explained by an LLM, and a narrow Joule Studio/MCP agent action with a named executing identity and human-in-the-loop confirmation. Closes with a governance and cost memo that keeps generative-AI-hub metering (tokens to GenAI tokens to capacity units, SAP Note 3437766) separate from AI Units, and a precise statement of what changes between a trial-tenant build and production. Calibrated to the depth expected by the C_AIG_2604 exam.

What you will learn

  • Assemble five previously learned SAP AI capabilities (semantic grounding source, vector retrieval, orchestration, tabular prediction, narrow agent) into one governed, end-to-end architecture
  • Explain and apply the generative AI hub orchestration pipeline's fixed execution order for a solution combining grounding, masking, filtering and a model call
  • Combine a tabular prediction (SAP-RPT or HANA PAL) with an LLM explanation step that never invents an uncomputed cause
  • Design a narrow, single-tool agent action with a named executing identity, exact input schema and a human-in-the-loop confirmation step
  • Produce a governance and cost close-out that keeps generative-AI-hub metering (tokens/GenAI tokens/capacity units) and AI Units separate, with no invented totals
  • State precisely what changes between a trial-tenant build and a production deployment of the same solution

Module overview

Who this is for. The final module of the SAP AI path (M333 through M384). It assumes you have built or specified the three portfolio demos (M382), can position the work honestly (M383), and can cite verified industry evidence (M384). This capstone does not introduce a new SAP capability. It assembles five capabilities you have already met — Datasphere or Business Data Cloud as a grounding source, HANA Cloud's Vector Engine, the generative AI hub's orchestration pipeline, a tabular prediction (SAP-RPT or PAL), and a narrow Joule/MCP agent action — into one coherent, governed, end-to-end solution, and names exactly what separates a trial-tenant build from a production one.

Prerequisites

  • Completion of M382, M383 and M384, or equivalent portfolio, positioning and evidence-discipline experience
  • Working knowledge of the generative AI hub orchestration service (M325), HANA Cloud Vector Engine (M330), SAP-RPT/PAL (M338-M340), and Joule Studio/MCP agents (M326, M327)
  • Access to a generative AI hub trial tenant, a HANA Cloud instance with the NLP feature enabled, and a Joule Studio sandbox is recommended but not required for the written exercises

Outcomes

  • Draw and defend a five-layer SAP AI architecture from memory, naming the exact SAP capability and its GA status at each layer.
  • Produce a complete, correctly ordered orchestration configuration combining grounding, masking and filtering for a real use case.
  • Deliver a governance and cost close-out that a client's security and finance teams could both sign off on.
  • State the precise, concrete gap between a trial-tenant build and a production deployment of the same solution.

Full module available to members. The full module adds: the decision framework · the end-to-end scenario walkthrough · the KPI scorecard · the anti-patterns · the code blocks · the knowledge check · the diagrams.

Open in the app →

Next in this track