SAP Business AI

What is SAP Business AI?

SAP Business AI is the collective name for the AI capabilities SAP builds into its own applications rather than sells as a separate product: Joule, the assistant embedded in the SAP work surfaces, plus the scenario-specific models in Sales, Service, Commerce and ERP. SAP’s Business AI overview is the canonical reference, and SAP’s AI documentation covers the platform side.

The distinction that matters commercially: availability depends on your release, your region and your licences, and it moves. Something demonstrated at a conference may not be in your tenant. Some functions come with the cloud subscription; others are licensed separately under your SAP entitlements, for example as AI units. We check the release, region, entitlements and available functions for your specific tenant before agreeing scope, and we keep roadmap items separate from what you can use now.

Spadoom implements SAP Business AI as part of Cloud ERP and Customer Experience delivery across Switzerland, Germany, Austria and Italy.

Start with a task you can measure.

We use AI in our delivery: analysis, code, tests and documentation, with our team responsible for the result. For your business, we begin with a concrete workflow and an agreed baseline. A demo shows an idea; a pilot tests it on authorised data; production requires acceptance, monitoring and a named owner.

In production today: OPO Oeschger

Customer project. OPO Oeschger, a Swiss B2B wholesaler, asked us to restart a stalled SAP Sales and Service Cloud V2 project. Phase 1 went live in three months, connected to S/4HANA, SAP Datasphere and Commerce Cloud, and with four AI use cases in daily operation:

• Intelligent visit planning prioritises which customer gets visited when, based on data from the whole customer base.
• AI ticket routing sends incoming service cases to the right team without manual triage.
• Business entity extraction picks the relevant information out of free text and documents and processes it in structured form.
• Intelligent contact impulses with SAP Datasphere spot unusual patterns in customer behaviour and trigger a follow-up.

These run in production. The next wave of AI use cases is planned for Phase 2 and is not live yet. The interactive examples elsewhere on this site are illustrations, not customer systems. Read the OPO Oeschger story.

Three places it earns its keep first

Data quality before migration. This is usually the highest-value starting point, because the data foundation is what every later AI feature inherits. It fits naturally into an ERP move, where master data gets cleaned anyway.

Service triage. In Service Cloud V2, classification and routing can be measured against the categories your team already uses. The OPO routing above is this pattern in production.

Field-service planning. In FSM, the value depends on how well skills, availability and travel data are maintained. We check that before promising better schedules.

Joule in Sales Cloud V2 and Service Cloud V2

In SAP Sales Cloud V2, Joule can summarise an account before a rep walks into a meeting, and the product offers AI support for forecasting and opportunity scoring. In Service Cloud V2, it supports case classification, response suggestions and knowledge recommendations. The features are real; adoption is a configuration and change question, and the gap between “activated” and “used” is where the work lies.

We describe the practical side in SAP Joule in Sales Cloud V2: a hands-on guide and in what agentic AI means for sales.

Joule agents and the integrated stack

Agentic scenarios are where the integrated SAP stack pays off. An agent that checks stock in S/4HANA Public Cloud before committing a delivery date, drafts the quote and logs the activity in Sales Cloud V2 works best when CRM and ERP share one business partner and one identity. On a fragmented landscape, the same scenario is an integration project before it is an AI project. Because we deliver the ERP core and the CX layer with one team, that integration has one owner.

Why AI projects stall at the data seam

AI does not remove integration work. ERP and CX must expose the required data through supported interfaces and permissions, which is SAP BTP work and, where the case justifies it, Business Data Cloud. A shared data platform is an architecture choice, not a prerequisite for every use case, and we will say so when a single interface would do.

A model trained on duplicate accounts learns the duplicates. A summary of a fragmented account history is a fragmented summary, fluently worded. Whatever quality sits underneath is what comes out, amplified.

From pilot to production

Engineers review generated code and own the integration. Before production, we agree who monitors the results, who handles errors and which costs are acceptable, including AI units consumed under your SAP entitlements. If the pilot does not clear the threshold, that is a result too, and a cheaper one than a production rollout nobody uses.

Which CX Joule assistants are available and which are only announced: our Joule assistant navigator lists each one with its SAP source.

Our own MCP server for SAP Sales and Service Cloud V2 makes CRM data available as tools for AI assistants.

What Spadoom delivers

• Joule configuration in Sales Cloud V2, Service Cloud V2 and S/4HANA Public Cloud, scoped to the processes where it measurably helps
• Adoption work: enablement, guardrails and the change work that turns activated features into used ones
• Agentic scenarios across CRM and ERP, built on the integrated SAP stack
• Custom AI extensions on SAP BTP where the standard stops
• The data foundation underneath, from a clean interface to SAP Business Data Cloud

Make the first step small

Bring one process that costs your team time. We define the decision, the data, the human review and the measurement. We do not present illustrative workflows as productive customer references.

Discuss an AI use case or see how this fits the wider transformation.