Content
Few SAP topics have attracted as much attention in recent months as the combination of SAP Business Data Cloud (BDC) and the AI assistant Joule. On stage and in presentations, everything already seems to be in place: you ask a question in natural language and receive a reliable answer based on your company’s data. In practice, however, the picture is more nuanced. This article explains what the two products can do, how they differ from traditional SAP analytics tools, which use cases are realistic and what companies should be aware of now.
What exactly are SAP Business Data Cloud and Joule?
What Is Behind SAP Business Data Cloud?
SAP introduced Business Data Cloud in February 2025 in partnership with Databricks. It is designed as a SaaS platform that brings together previously separate components under one roof: SAP Datasphere, SAP Analytics Cloud, selected SAP Business Warehouse capabilities and a managed version of Databricks (SAP Databricks) for data science and machine learning. In addition, it includes pre-built data products for SAP business applications and so-called Intelligent Applications that build on these data products.
The benefits go beyond the technology itself: data products and a data catalogue can give fresh impetus to the often-neglected topic of data ownership (data governance, data mesh). This can improve data maturity and, in turn, the quality of insights generated by AI. Furthermore, business departments and IT are more likely to discuss data products rather than InfoObjects, fields or attributes.
The underlying concept is a “Business Data Fabric”: data should remain as close as possible to where it originates and be made available together with its business meaning (semantics). A data product called “Invoices”, for example, therefore knows not only about tables and fields, but also about relationships, hierarchies and business logic. BDC Connect allows external platforms such as Databricks or Google Cloud to be connected using a zero-copy approach, meaning that data does not need to be copied. Further integrations, for example with Snowflake, are already possible, while integration with Microsoft Fabric has been announced.
A data fabric is characterised by unified and democratised access to data: data can be distributed across different locations, access should be available in real time, and integration should be largely automated. Redundant data storage, transformations persisted across multiple layers and the physical integration of different sources can all stand in the way of this approach. Compared with SAP BW, BDC is much more closely aligned with this concept and more open to the wider market, while data security and authorisations remain manageable.

Checklist "Fom SAP BW or SAP BW/4HANA to SAP Datasphere"
Our checklist helps you find the right migration path from SAP BW or BW/4HANA to SAP Datasphere.
Download
What is Joule and what Role does AI play?
Joule is SAP’s generative AI assistant. It was announced in 2023 and is now embedded in numerous SAP solutions, ranging from S/4HANA Private Cloud and Public Cloud to SuccessFactors, Ariba and the Business Technology Platform. Users can ask questions in natural language, have relationships and connections explained to them or trigger actions within the respective system.
Since 2025, SAP has been taking things a step further: Joule is intended not only to provide answers but also to handle multi-step tasks through Joule Agents, such as comparing quotations, resolving disputes or preparing enquiries. With Joule Studio, companies can also build their own agents using low-code tools. The role of BDC is crucial here, as it provides the data context. Without clean data enriched with semantic information, AI remains a well-spoken generalist that does not understand your business.
Joule is part of SAP Business AI, the umbrella term for SAP’s AI portfolio, which comprises three layers: Embedded AI (AI scenarios within individual products, such as classification, forecasting or automation steps), Joule as a unified natural language interface across applications, and the AI Foundation on SAP BTP as the underlying technical platform, including governance and security (e.g. Generative AI Hub, SAP Document AI and Joule Studio). The overall vision is AI that accelerates processes in a controlled manner, taking SAP process logic and company data into account.
How does it all fit into SAP's Strategy?
SAP’s strategy can be summarised in three key areas:
- Cloud first: Many new AI features are tied to cloud offerings such as RISE with SAP or GROW with SAP. Joule therefore also serves as an argument in favour of cloud migration.
- Data as the foundation: BDC is intended to provide the foundation for all AI applications within the SAP ecosystem.
- Openness with strategic interests: Through partnerships with Databricks, Google Cloud and other providers, SAP is opening up its data to external platforms while seeking to retain semantic control over SAP data.
BDC is positioned as a strategic platform rather than an isolated tool. The aim is to bring together data silos and separate solutions such as data warehouses, data lakes and standalone BI or AI tools.
How does Joule differ from SAP Analytics Tools?
Traditional SAP analytics tools such as BW, BusinessObjects or SAP Analytics Cloud are based on predefined models, reports and dashboards. Business users consume what IT or the BI team has built in advance. Joule shifts this approach in four key ways:
- User interaction: Instead of setting filters and navigating reports, users simply ask a question. This lowers the barrier to entry, including for people without BI expertise.
- Context rather than reports: Joule combines data from multiple applications and can explain results (e.g. “Why has cash flow declined this quarter?”) rather than simply displaying figures.
- Action rather than display: With agents, the response can extend into the business process itself. Joule can suggest or execute follow-up steps, whereas traditional analytics tools typically stop at the dashboard.
- Flexibility: Not every question needs to be modelled in advance. This is an advantage when it comes to ad hoc analysis.
It is important to understand that Joule does not replace existing tools. Clearly defined and validated reports remain essential, particularly for regulatory reporting, consolidated financial statements and planning. Joule provides an additional access layer; it is not a substitute for a well-designed data model.
Which Use Cases can already be implemented with SAP BDC & Joule?
Realistically, the most viable scenarios today are those in which data is already structured and stored in SAP systems and the questions being asked are clearly defined. The following examples illustrate what such scenarios might look like. Whether a particular capability is available as a standard feature or requires custom configuration or development should be checked against SAP’s current roadmap on a case-by-case basis.
The Intelligent Applications provided by BDC offer a useful starting point. They are not monolithic solutions but are tailored to individual business areas, including:
- People Intelligence: Modern HR analytics based on SuccessFactors data, divided into Workforce Composition Insights, Skill Insights and Compensation Insights.
- Cloud ERP Intelligence: For SAP S/4HANA Cloud Private Edition, currently featuring 195 data products covering accounting, sales, logistics and manufacturing. The SAC story “Working Capital” is available as a user interface.
- Finance Intelligence: For SAP S/4HANA Cloud Public Edition, with 150 primary data products, currently without visualisations.
- Supply Chain Intelligence: Building a resilient supply chain through intelligent applications that connect planning, procurement, logistics and operations.
- Spend Intelligence: Providing full transparency across business activities through clean, connected data, breaking down silos to generate insights and drive results.
- Revenue Intelligence: Bringing together data from sales, service, marketing, commerce and ERP while preserving semantics and governance.
CFO: Cashflow-Insight
The finance department wants to understand how liquidity is developing and why. BDC can bring together data from accounts receivable, accounts payable, bank accounts and planning. Joule can answer questions such as “Which ten customers are responsible for the longest payment delays?” or “How would liquidity change if suppliers shortened their payment terms?”. The added value lies in rapid root cause analysis and the simulation of straightforward scenarios. However, the validated model remains authoritative for official liquidity planning. Cloud ERP Intelligence provides a concrete starting point: the SAC story “Working Capital” illustrates developments in net working capital and the cash conversion cycle, while monitoring key figures such as payment terms, days sales outstanding and inventory levels. Finance Intelligence is also intended to provide planning and forecasting capabilities, as well as AI-powered anomaly detection.
Procurement: Spend Intelligence
In procurement, transparency over expenditure is key. How are purchasing volumes spread across too many suppliers? Which product categories show price discrepancies? Where are framework agreements being bypassed (“maverick buying”)? Joule can analyse expenditure by category, supplier or location and explain any anomalies. Agents in Ariba already support quotation analysis, for example by comparing supplier bids based on total cost. A clean material group and supplier structure is a prerequisite. The Spend Intelligence Intelligent Application builds on Ariba data. It is designed to consolidate expenditure across systems using consistent categories, classify invoice line items through machine learning and identify anomalies such as duplicate spending or uncompetitive prices.
CHRO: Talent Flow Overview
HR management is interested in employee arrivals and departures, internal transfers and bottlenecks in critical roles. By combining data from SuccessFactors and other sources, it is possible to answer questions such as “In which areas is employee turnover above average?” or “How many key positions are expected to become vacant over the next 24 months?”. Here, in particular, data protection, works council involvement and the authorisation concept must be clarified before implementation. Analyses at individual employee level are sensitive. People Intelligence already provides an available Intelligent Application for this purpose.
Attribution and Forecasting
In sales and marketing, the focus is on attributing revenue to influencing factors (attribution) and predicting future developments (forecasting). SAP Databricks can be used to train predictive models on semantically enriched SAP data, for example to forecast sales volumes or demand. Joule then serves as the interface for querying and explaining the results. It is important to note that forecast accuracy depends less on Joule itself than on historical data and model quality. Revenue Intelligence (from quotation to payment receipt) and Customer Intelligence (covering the entire customer journey) have been announced for this area.
Further Examples
- Supply Chain: Early warning of potential supply bottlenecks by linking inventory, order and supplier data. Supply Chain Intelligence (announced) is intended to additionally combine SAP data with external sources such as weather data, government information, market data and information from the SAP Business Network.
- Order Management: Converting unstructured documents, such as orders received by email, into transaction-ready data.
- Service and Claims: Accelerating case handling by having agents gather relevant information in advance and make recommendations.
- Controlling: Variance analysis with automatically generated explanations for management reporting.
What are the Limitations of SAP Business Data Cloud & Joule?
What does not (yet) work as promised?
SAP has expanded its AI portfolio very rapidly, claiming to now offer more than a thousand pre-built skills and dozens of agents. However, reality in businesses is lagging behind. According to the DSAG Investment Survey 2026, only a small single-digit percentage of the SAP customers surveyed are currently using Business AI productively. Independent assessments also describe Joule as useful but in need of improvement, pointing out that some features are still under development.
The general rule is: AI agents do not understand the data they process in the human sense; instead, they work with probabilities. They can produce plausible but incorrect results (hallucinations), are not automatically audit-proof and are vulnerable to prompt injection because they cannot reliably distinguish legitimate requests from hidden malicious instructions. They cannot assume organisational or legal responsibility either. They therefore cannot replace business departments, internal control systems or management decisions.
In practice, agents with clearly defined tasks, verifiable results and integration into existing role and authorisation concepts are therefore the ones most likely to succeed. Across industries, AI agents are already integrated into customer service in 56 per cent of companies and into operations in 39 per cent. Trust in AI is highest when it comes to data analysis and generating insights.
Typical gaps between promises and practical implementation:
- Maturity: Announcements, preview features and generally available features are often mixed together in communications. Check the actual availability status for every use case.
- Response quality: Generative systems can produce incorrect or inaccurate answers. This is unacceptable when dealing with financial figures unless the results are validated.
- Traceability: It must be clear which data a response is based on. Without this transparency, acceptance among controlling and internal audit teams will be lacking.
- Access: As things currently stand, many features are tied to cloud contracts. On-premise landscapes are partly excluded or can only be connected via workarounds. Language limitations (e.g. fallback to English) and phased rollouts are additional factors. Joule features are often dependent on specific use cases, modules or programmes, and Joule is not consistently available as an assistant in some analytics scenarios. A unified agent layer covering all SAP interfaces, system variants and processes does not yet exist.
Where are manual Interventions still necessary?
- Functional validation of results, particularly for analyses involving financial data or personal information.
- Modelling and maintenance of data products, semantics and authorisations.
- Configuration, approval and monitoring of agents, including clear rules defining what an agent may do autonomously and where human approval is required.
- Adjustments for company-specific processes and KPIs that do not correspond to the SAP standard.
Joule is currently evolving at a rapid pace, meaning that the points listed here may need to be revised even over relatively short periods of time.
What Challenges exist in Terms of Data Quality and Integration?
The most common reason why AI projects fail to meet expectations is not the model itself, but the underlying data. Typical problems include:
- Fragmented system landscapes: Multiple ERP systems, legacy systems and third-party applications using different definitions for the same KPI.
- Master data quality: Duplicate records, incomplete supplier and customer data or inconsistent material groups distort analysis results.
- Custom developments: Heavily customised systems (Z-tables, bespoke logic) are difficult to convert into standardised data products.
- Non-SAP data: Connecting external sources is possible, but requires considerable effort. The promised openness needs to be assessed for each individual data source.
- Governance: Who is allowed to see which data, and which data may Joule use as a basis for its responses? Without clear answers, compliance risks arise. A gradual, non-invasive approach based on building a data culture is more useful than a 100-page policy document. The data catalogue promotes reusability and trust.
- Data literacy: Data without context does not provide information, and without data literacy, analysing it only leads to good decisions by chance. Calls for an “export to Excel” button should be seen as a warning sign. The minimum level of maturity for a modern data culture has been reached when analysis takes place where the data is made available.
- Costs and licensing: Consumption-based pricing models and additional components make overall costs difficult to plan.
Who would benefit from using SAP BDC & Joule and who would not (yet)?
Which Companies benefit the most?
The greatest benefits are likely to be seen by companies already working with SAP cloud solutions (e.g. S/4HANA Private Cloud, SuccessFactors, Ariba, Datasphere), following a largely standard-based process approach and having a sound data foundation. Organisations with multiple SAP systems that have previously struggled to consolidate data can also benefit from a central data foundation enriched with semantic information. Traditional BW customers can carry forward the value of their existing investments to a modern platform and enhance it with AI. Complex integration scenarios involving multiple analytics solutions and redundant data storage can be eliminated.
In which scenarios does Joule deliver real added value?
- Recurring questions that previously required a report or a request to IT.
- Root cause analysis and explanations of variances rather than simply displaying figures.
- Processes involving a high level of manual preparation, such as comparing quotations, capturing documents or preparing cases.
- Business areas where specialist departments have so far had very limited access to data.
When should its use still be viewed with caution?
- In heavily customised on-premise landscapes without a concrete cloud roadmap.
- Where data quality is unclear and data ownership has not been established. AI tends to amplify existing data problems rather than solve them.
- In highly regulated processes where results must be fully traceable and non-deterministic responses represent a risk.
- Where there is no clear business case and implementation is driven purely by technology.
- Where the resources required for governance, training and operations are lacking.
What should Companies do now?
How can you prepare stategically?
The most important strategic decision is to treat data as a topic in its own right, rather than as an afterthought in an AI project. This includes aligning the SAP roadmap (cloud migration, RISE/GROW) with the data strategy, defining responsibilities for data products and establishing clear guidelines for the use of AI: What can be automated, what cannot, and who is accountable? This also involves building data literacy and fostering a data culture. Both take time to develop and require the right methodology and mindset. Key elements include establishing a common language (company-wide definitions of terms, data catalogue) and creating a learning culture through mandatory basic training and gamification approaches such as badges and levels. Ideally, these initiatives should be embedded in a campaign supported by the executive board. Experience shows that including data-related objectives in employee performance reviews is not enough.
What are the first steps worth taking?
- Take stock: Which systems, data sources and reports are in place? Where are the biggest data quality issues?
- Prioritise use cases: Select two or three specific scenarios with measurable benefits rather than trying to tackle everything at once.
- Check availability: For each use case, clarify which components are generally available, what needs to be licensed and whether your own system landscape meets the requirements.
- Establish data maturity: Set up data products, semantics, master data cleansing and an authorisation concept for the selected use cases.
- Launch a pilot: Start with a small group of users, systematically evaluate response quality and collect feedback.
What should you consider when implementing the solution?
- Governance first: Involve data protection, the works council, internal audit and IT security at an early stage.
- Bring people on board: Provide training and manage expectations. Joule is an assistant, not an oracle.
- Measure quality: Track accuracy, time savings and user acceptance using clearly defined KPIs.
- Human in the loop: Critical decisions and postings must remain subject to clearly defined approval processes.
- Keep costs under control: Plan licensing, consumption-based and operating costs transparently from the outset.
- Monitor the roadmap: SAP is developing its solutions very rapidly. Review decisions regularly in light of the latest developments.
Conclusion
SAP Business Data Cloud and Joule represent a significant step towards data-driven, AI-supported business processes. The technology is no longer science fiction, but it is not yet a ready-made solution that delivers results without further effort. Companies stand to gain the most by getting their data foundation in order, focusing on a small number of carefully selected use cases and supporting implementation with governance and change management. By contrast, companies that rely solely on the promise of “ask a question, get an answer” without putting the necessary foundations in place are more likely to be disappointed.
Contact us!
Do you have further questions? Our experts will be happy to help.
Schedule a web meeting with us — or simply leave your question in the comment section.

You are currently viewing a placeholder content from Active Campaign. To access the actual content, click the button below. Please note that doing so will share data with third-party providers.
More Information
