BTP & Integrationbeginner

Generative AI Hub Part 1: SAP AI Core, Generative AI Hub and How They Relate

Key takeaway

SAP AI Core is the runtime and lifecycle foundation for AI on BTP; the generative AI hub is a capability within it (plus SAP AI Launchpad) for accessing and orchestrating LLMs. Learn the relationship between AI Core, AI Launchpad and generative AI hub, the extended service plan requirement, resource groups, and the models available (third-party and SAP's own, like SAP-RPT-1.

What this complete series covers

Learn Generative AI Hub and SAP AI Core in a clear, practical sequence — from the basic idea to real project use, troubleshooting, and interview-ready understanding.

Who it is for: Best for SAP consultants, developers, and learners who want a practical understanding of Generative AI Hub and SAP AI Core.
Prerequisites: A basic understanding of SAP BTP helps. Hands-on steps may require a BTP subaccount and access to the relevant Integration Suite or BTP service.
Final outcome: By the end, you should be able to explain Generative AI Hub and SAP AI Core clearly, follow the main end-to-end flow, apply it in a project scenario, and troubleshoot the most common issues.

Part 1: Generative AI Hub Part 1: SAP AI Core, Generative AI Hub and How They Relate
Understand how SAP AI Core and the generative AI hub provide governed access to foundation models, orchestration and enterprise AI capabilities on SAP BTP.
Part 2: Generative AI Hub Part 2: Deployments, Prompts and Consuming Models
Understand SAP's generative AI hub as a governed access and orchestration layer on SAP AI Core for enterprise language-model use cases, prompt development and model lifecycle management.
Part 3: Generative AI Hub Part 3: Production Patterns — Grounding, Data Masking, Content Filtering and Evaluation
Build production-grade generative AI with grounded enterprise context, deterministic tools, systematic evaluation, privacy controls and observable failure handling rather than prompt-only demos.

System and release assumptions: SAP BTP services evolve regularly. Capability names, screens, and setup steps can differ by region, entitlement, and tenant version; the core design principles remain the same.

Explanation

📌 KEY POINTS
✅ SAP AI Core provides runtime, lifecycle management and operational controls; generative AI hub extends it for LLM access and orchestration — it doesn't replace AI Core.
✅ Generative AI hub is available only with the AI Core extended service plan (upgrading from free/standard incurs cost).
✅ Resource groups organize deployments and quota; a default deployment exists for orchestration once the plan is active.
✅ Models include general-purpose LLMs (from providers like Azure OpenAI, Anthropic, and open source) plus SAP's own models (SAP-RPT-1.5 for predictive business insights, SAP-ABAP-1 for ABAP development assistance).
⚠️ SAP AI Core has no on-premise option — the generative AI hub logic always runs on BTP, even for scenarios reading on-premise S/4HANA data via principal propagation.

━━━━━━━━━━━━━━━━━━━━
▶ WHAT THIS SERIES COVERS
Part 1: How AI Core and generative AI hub relate
Part 2: Fundamentals — deployments, prompts and consuming models
Part 3: Production patterns — grounding, orchestration, evaluation

━━━━━━━━━━━━━━━━━━━━
▶ 1. FOUNDATION

SAP AI Core: runtime, security, scaling.

━━━━━━━━━━━━━━━━━━━━
▶ 2. GENERATIVE AI HUB

LLM access and orchestration on top of AI Core.

━━━━━━━━━━━━━━━━━━━━
▶ 3. SERVICE PLAN

Extended plan required; entitlement and quota.

━━━━━━━━━━━━━━━━━━━━
▶ 4. RESOURCE GROUPS

Isolation; deployments.

━━━━━━━━━━━━━━━━━━━━
▶ 5. MODELS

Third-party and SAP proprietary models.

━━━━━━━━━━━━━━━━━━━━
▶ STEP BY STEP

Check your subaccount's AI Core plan (free/standard vs extended).
Review resource groups and entitlements.
Find the default orchestration deployment.

✅ CHECKPOINT: you can explain why generative AI hub needs the extended plan.

Code example

ABAP Code

===== RELATIONSHIP =====SAP AI Core (runtime, lifecycle, security) → Generative AI Hub (LLM access + orchestration) → SAP AI Launchpad (UI: playground, deployments) ===== SERVICE KEY FIELDS (from AI Core instance) =====url (token endpoint) | clientid | clientsecret | AI_API_URL ===== TOKEN REQUEST =====POST {url}/oauth/token  grant_type=client_credentials&client_id={clientid}&client_secret={clientsecret}→ access_token used as Bearer token against AI_API_URL

Real project scenario

A team assumed generative AI hub was a separate subscription and delayed a pilot for weeks waiting on procurement. Once they realized it was an AI Core extended-plan capability, upgrading the existing AI Core instance unlocked it the same day.

Common mistakes

Assuming a separate product/subscription.
Skipping the extended plan requirement.
Ignoring resource group quotas.
Treating SAP proprietary models the same as general LLMs without checking fit.
Not verifying current plan names before a project.

Best practices

Confirm AI Core plan tier first.
Understand resource groups before deploying.
Check SAP Help Portal for current model list.
Distinguish SAP-RPT-1.5/SAP-ABAP-1 use cases from general LLMs.
Budget for the extended plan upgrade cost.

Interview angle

Questions: "What is the relationship between AI Core and generative AI hub?", "Why is the extended plan required?", "Which SAP proprietary models exist?"
Strong answer: AI Core as foundation, generative AI hub as extension, extended plan, resource groups, model landscape.

Practice this lesson

Quick check

1. Which SAP proprietary model is designed for predictive business insights across finance, sales and supply chain?

2. Which AI Core service plan is required to use the generative AI hub?

3. What is the relationship between SAP AI Core and the generative AI hub?

Interview questions

1. What are the most important concepts in Generative AI Hub and SAP AI Core, and how do they connect to Generative AI Hub and SAP AI Core Fundamentals in a real SAP landscape?

Answer it in your own words first, then compare.

2. A production issue is reported in Generative AI Hub and SAP AI Core around Getting Started with Generative AI Hub and SAP AI Core. How would you troubleshoot it systematically and avoid the common failure patterns?