SAP AI Core & AI Launchpad

What you will learn

• Explain what AI Core actually runs — inference, training orchestration, GenAI Hub proxies, model registry — and why AI Launchpad is the governance cockpit, not just a UI
• Apply AI Core in a typical SAP analytics engagement
• Recognize the 3-5 common mistakes and how to avoid them
• Position this skill in your personal brand and rate conversation

Module overview

SAP AI Core is the runtime; AI Launchpad is the cockpit. AI Core hosts and serves models — both SAP-built (Joule, embedding services) and bring-your-own (custom Hugging Face models, OpenAI proxies, Anthropic proxies). AI Launchpad is the BTP UI for managing deployments, monitoring, and governance.

The 4 things AI Core actually does.

1. Inference serving — REST endpoints for model calls (chat, completion, embedding, classification).
2. Training orchestration — fine-tuning + RAG pipeline runs against governed data.
3. Generative AI Hub — pre-built proxies to OpenAI, Anthropic, AWS Bedrock, Google Vertex (with SAP-managed billing + audit).
4. Model registry + versioning — every model deployment versioned with metadata for Art. 11 dossier.

Why a senior consultant cares. Without AI Core + Launchpad, every AI feature is a one-off integration with no governance. With them, inference is a capability the platform exposes, deployments are auditable, costs are attributable per business unit, and EU AI Act Art. 11 documentation flows naturally from the registry.

Prerequisites

• Intermediate hands-on experience on SAP analytics projects
• Review core concepts first: C033, C026, C008

Outcomes

• Map AI Core's four core capabilities (inference, training, GenAI Hub, registry) to a governed deployment architecture
• Apply AI Core in a typical SAP analytics engagement
• Explain the core architecture and decision points for SAP AI Core & AI Launchpad
• Apply a repeatable implementation pattern in a 15-minute lab format