Concept card

SAC Smart Insights and Smart Discovery — and What Just Ask Changes

As of 2026-09-25

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 SAC Smart Insights and Smart Discovery?

Smart Insights (automatic explanation of a selected chart point) and Smart Discovery (automated correlation/segment scan across a dataset) are classical, pre-Joule SAC features — SAP documents them as classical ML/statistics, not generative AI. Just Ask adds a third, natural-language entry point into the same indexed models; SAP does not document a technical dependency between Just Ask and the two classical features, and answer quality for all three depends on the same semantic-modeling fundamentals a consultant already owns.

Two entry points that predate Joule

Smart Insights and Smart Discovery are two of the four components SAP groups under SAC's "Smart" augmented-analytics umbrella (alongside Smart Predict, concept C341, and Search-to-Insight). Smart Insights generates an automatic explanation for a data point a user selects in a chart — contributors, trends and outliers behind that specific number — while Smart Discovery runs an automated analysis across an entire dataset to surface the strongest correlations and segments without a user selecting anything first. Both predate Joule by years, and both are classical statistics and machine learning, not generative AI — SAP's own current documentation for Joule in SAC states this plainly: Smart Insights and Smart Predict remain in the same help chapter as Joule's generative features but are classical ML and statistics, not generative AI themselves. That sentence is the one to hold onto when a client conflates "AI in SAC" with "generative AI in SAC" — Smart Insights predates the LLM era of SAC entirely and would keep working even if every generative feature in the product were switched off.

Why it matters

SAP's own current documentation states Smart Insights and Smart Predict are classical ML/statistics, not generative AI, despite sharing a help chapter with Joule — a client conflating 'AI in SAC' with 'generative AI in SAC' needs this distinction made explicit, not assumed.
SAP does not document Just Ask as calling Smart Insights/Smart Discovery under the hood, and neither classical feature is documented as deprecated — describing Just Ask as 'Smart Insights with a chat interface' overstates an undocumented technical relationship.
What genuinely changed is that answer quality now depends on semantic modeling (synonyms, date fields, measure typing) that a consultant already owns as a skill — reframing 'AI accuracy' as a modeling deliverable rather than a black-box tuning problem.

Key points

Smart Insights and Smart Discovery are two of SAC's four 'Smart' augmented-analytics components, alongside Smart Predict (C341) and Search-to-Insight.
Smart Insights: automatic explanation (contributors, trends, outliers) for a user-selected chart data point.
Smart Discovery: automated scan across a whole dataset surfacing correlations and segments, no user selection required first.
SAP documents both as classical ML/statistics — explicitly not generative AI — even though they share a help chapter with Joule's generative features.
Just Ask is a separate, natural-language query engine, part of Joule's footprint in SAC (C335) — a third entry point into the same indexed SAC/Datasphere models.
SAP's material does not describe Just Ask as calling Smart Insights/Smart Discovery internally, nor either as deprecated in favor of Just Ask — coexistence, not replacement.
SAP documents Just Ask answer quality as governed by admin settings (synonyms, default measures/filters, When/How rules) and a readiness check flagging semantic issues (dates, duplicates, IDs-as-measures, currency).
The practical implication: a semantic-modeling fix tends to improve Just Ask and the classical Smart features simultaneously, since they read the same underlying models.

Terms used on this page

Smart Insights: A SAC feature generating an automatic statistical explanation (contributors, trends, outliers) for a chart data point the user selects.
Smart Discovery: A SAC feature running an automated scan across a full dataset to surface strong correlations and segments without prior user selection.
Just Ask: SAC's natural-language query engine, part of Joule's footprint, searching indexed model metadata to answer typed questions.
Readiness check: SAP's documented tool flagging semantic-modeling issues (date modeling, duplicates, IDs as measures, currency handling) that degrade Just Ask answer quality.
When/How rules: Administrator-configured Just Ask settings controlling when and how specific answers or behaviors are triggered.

Sources

Full card available to members. What the full card adds: the full decision framework · the common pitfalls and their fix · the cheat sheet · the facts worth quoting.

Open in the app →