Concept card

Knowledge Core — The Data-Fabric-for-AI Evolution

As of 2026-07-24T14:00:00Z

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

What is Knowledge Core — The Data-Fabric-for-AI Evolution?

Knowledge Core is SAP's answer to why raw data isn't enough for agentic AI — six capabilities (semantics, knowledge graph, AI domain models, analytics, data products, discovery) sit as a governed layer between raw data and AI agents.

Knowledge Core is the architectural layer SAP positioned at Sapphire 2026 as the piece every organisation is missing before agentic AI can be trusted to act on its data. The framing SAP gave the keynote audience is a three-step evolution of enterprise data architecture. The data warehouse generation was built for a deterministic world, with compute, the operational database and storage all tightly coupled together. The lakehouse generation separated compute from storage and unified reporting, machine learning and data engineering on a single lake. Knowledge Core represents the third step: a data fabric built specifically for the AI era, on the premise that raw data alone is not enough — an autonomous agent needs a typed, governed, semantically rich layer to reason over before it can be trusted with a business decision.

What it actually contains

Why it matters

The evolution framing is explicit: Data Warehouse (coupled compute/storage) → Lakehouse (separated compute/storage) → Data Fabric with Knowledge Core (typed, governed substrate agents can reason over) — each step solves a problem the last one didn't.
Semantics is named as the specific layer without which "an agent cannot generate a trusted answer" — the business-definition gap is the stated failure mode, not a generic governance concern.
The knowledge graph capability is what lets an agent traverse sales order → fulfilment → finance without hand-coded joins — a concrete architectural payoff, not an abstract claim.

Key points

Three-step evolution per Sapphire 2026: Data Warehouse (tightly coupled) → Lakehouse (compute/storage separated) → Data Fabric with Knowledge Core (AI-era semantic + governed layer).
Six Knowledge Core capabilities: Semantics · Knowledge Graph · AI Domain Models · Analytics & Simulations · Data Products · Discovery & Activation — each a distinct, non-overlapping functional block.
Three orchestration wrappers: FinOps & Lifecycle Management · Data Governance (MDM, DQ, stewardship, DataOps, catalog, lineage) · Security & Trust.
Stack position: Lake Storage (SAP data + BW data + non-SAP data) → Knowledge Core → AI Agents & Applications (Joule Agents + Intelligent Applications) — three distinct tiers, not two.
SAP One Domain Model underpins the Data Governance wrapper: it provides the canonical business object definitions (customer, material, cost centre, plant) that give Knowledge Core its semantic backbone across SAP applications.
Knowledge Core vs Lakehouse: a lakehouse separates compute from storage but still serves raw tables; Knowledge Core adds the semantic, graph, governance, and AI model layer that makes raw data agent-consumable.
Knowledge Core vs KMS (Knowledge Management System): KMS was document-based (SharePoint-class); Knowledge Core is structured data + active metadata + graph + AI models in a single governed plane.
Discovery and Activation is the capability that closes the loop from user query to governed answer to triggered action — it is the 'last mile' of the Knowledge Core that makes the stack business-usable, not just technically complete.
Data Products within Knowledge Core are versioned, governed, and carry SLAs — Finance Product, Supply Chain Product, Spend Intelligence Product each have defined consumers and update contracts.
When to use Knowledge Core framing: any BDC architecture conversation, any 'why isn't our lakehouse enough for AI?' objection, any Joule extension scoping discussion — this is the 2026 SAP architecture narrative.

Terms used on this page

Knowledge Core: The Sapphire 2026-introduced central architectural layer in SAP's Data Fabric model; combines Semantics, Knowledge Graph, AI Domain Models, Analytics & Simulations, Data Products, and Discovery & Activation under unified governance to make enterprise data agent-consumable.
Data Fabric: The umbrella AI-era architectural pattern where data is woven across SAP and non-SAP sources via a Knowledge Core layer; distinct from the Lakehouse pattern (which lacks the semantic/governance/graph tier required for autonomous AI agents).
SAP One Domain Model: SAP's canonical ontology of business objects — customer, supplier, material, plant, cost centre, project — that provides the semantic backbone for Knowledge Core's Semantics and Data Governance capabilities; ensures Joule agents use the same business definitions across S/4HANA, Ariba, SuccessFactors, and third-party data.
AI Domain Models: Pre-built predictive and prescriptive models within Knowledge Core targeted at specific business domains (financial close forecasting, supply chain risk scoring, customer churn prediction) and industries; reduces the AI build burden for organisations without data science teams.
Data Products: Curated, versioned, governed data assets within Knowledge Core shaped for downstream consumption with defined SLAs and consumer contracts; examples include Finance Product, Supply Chain Product, Spend Intelligence Product.
Discovery and Activation: The sixth Knowledge Core capability that routes a user or agent query through the semantic and graph layers to the right data product and triggers actions; the 'last mile' capability that makes Knowledge Core business-usable rather than technically complete.
Active metadata: Metadata that participates in query execution and governance enforcement at runtime (as opposed to passive catalog metadata that is only read by humans); part of Knowledge Core's Semantics capability; enables automated data quality checks, lineage propagation, and semantic query rewriting.
DataOps: Operational discipline within the Data Governance wrapper of Knowledge Core that applies DevOps practices — version control, testing, CI/CD pipelines — to data pipeline development and deployment; enables reliable, traceable updates to data products.

Sources

Full card available to members. What the full card adds: the full decision framework · the SAP vs Snowflake / Databricks / Fabric comparison · the common pitfalls and their fix · the cheat sheet · the architecture schemas · the code blocks · the facts worth quoting.

Open in the app →

Guides that answer with this page

These guides cite this page as one of the sources their answer rests on.