AI & Analytics Legends The knowledge platform for SAP Analytics
Academy module

Joule AI Copilot

Joule grounding flow: business question through governed data products to a provenanced answer, contrasted with an ungrounded generic LLM — architecture diagram for Joule AI Copilot, Analytics Legends Academy module M046

As of 2026-10-10

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

Joule is SAP's generative-AI assistant (announced 26 September 2023) and, with Joule Work presented at Sapphire 2026, the primary way users interact with SAP. For analytics, SAP's reference architecture is explicit: Joule calls SAC's JustAsk service, which searches indexed model metadata in a customer-specific vector store, picks an analytic model, has an LLM generate an OData/SQL query and runs it on Datasphere or HANA Cloud — honouring the user's SAC roles via principal propagation. So answer quality is decided by which models you expose, how they are named and described, and whether authorisations are right. Commercially, Joule Base is included with qualifying cloud subscriptions; from Q3 2026 Premium AI centres on agents consuming AI Units (0.02 per agent action in 2026). Measure per engagement — no borrowed accuracy figures.

What you will learn

  • Map the 2026 Joule portfolio — Joule, Joule Agents and Assistants, Joule Studio (designer, code editor, CLI), Joule Work, Joule for Consultants/Developers — and identify which one a client actually needs
  • Explain, from SAP's Analytical Insights reference architecture, how Joule turns a question into a governed query (JustAsk, vector search over model metadata, OData/SQL generation, Datasphere/HANA Cloud, principal propagation)
  • Design what Joule can see: the exposed model set, business naming and descriptions, single measure definitions, and an authorisation review of every exposed model
  • Scope the commercial boundary correctly: Joule Base at no additional cost versus Premium AI agents consuming AI Units from Q3 2026, without quoting unsourced prices
  • Measure answer quality per engagement with a dated evaluation set that scores model choice, query correctness and figure correctness separately

Module overview

Joule is SAP's generative-AI assistant, and in 2026 it is becoming the main way users work with SAP. Announced on 26 September 2023 and first shipped with SAP SuccessFactors and SAP Start at the end of 2023 (SAP S/4HANA Cloud Public Edition followed in early 2024), Joule has grown from a chat panel on top of existing screens into what SAP presented at Sapphire 2026 as the primary interface: with Joule Work, "users will now interact primarily with Joule" instead of navigating individual applications. For an analytics consultant the question is narrower and more useful: what does Joule do with the customer's governed data, how does it do it, what does it need to be trusted, and what does it cost?

Prerequisites

  • Working knowledge of SAP Analytics Cloud models and roles, and of Datasphere or BDC analytic models
  • Review core concepts first: C024, C027, C101

Outcomes

  • Explain to a board, accurately, where a Joule figure comes from and which controls apply to it.
  • Produce an exposure design for Joule: models, naming standard, authorisation review, change-calendar check.
  • Write a scope note separating Joule Base from AI Unit consumption, sourced to SAP's commercial-model documentation.
  • Run a small, dated evaluation set that tells you whether errors come from model choice, query generation or data freshness.

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