SAP Sapphire 2025 marked a strategic move beyond adding chatbots to business software: SAP wants Joule and specialized agents to use governed business data and work across operational processes. The direction is significant, but the announcements did not mean every capability was production-ready or available to every customer. For SAP customers, the practical question is whether their cloud systems, data, processes, contracts and controls are ready for AI to do more than answer questions.
What changed at SAP Sapphire 2025?
At its May 2025 event, SAP described a model in which AI is connected to business applications, data and workflows, rather than confined to a chat window inside one product. In the earlier copilot pattern, a user asks for an answer, summary or draft. SAP’s newer ambition is for AI to understand business context, recommend a response and, where authorized, help move work forward across applications.
SAP framed this as a “flywheel” linking applications, data and AI. In practice, the proposed pieces are SAP’s operational applications, Business Data Cloud, Knowledge Graph, Joule, specialized Joule Agents and SAP Business Technology Platform (BTP). The strategy matters because SAP software is often where decisions become orders, postings, approvals and other transactions. But connecting these pieces in a customer’s landscape is a technical and commercial undertaking, not an automatic consequence of attending the event.
SAP said its AI innovations could produce productivity gains of up to 30%. That is SAP’s claim, not an independently verified average or a guaranteed result for customers. SAP’s Sapphire announcement sets out the company’s positioning.
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11#1 Best Overall
How the proposed architecture fits together
| Layer | Role in SAP’s strategy | What to check in practice |
|---|---|---|
| Business applications | Hold operational workflows and transactional authority, such as finance, procurement, HR and supply chain. | Which applications and cloud editions are in use, and whether the workflow is standardized enough to automate. |
| Business Data Cloud | Provide governed data products and access to SAP and external data for analytics and intelligent applications. | Whether relevant sources are connected, definitions are consistent and the required data products are available to the customer. |
| Knowledge Graph | Represent relationships among business entities so AI can use context, not just retrieve text. | Whether relationships and metadata are modeled accurately and maintained. |
| Joule | Act as the conversational and task-oriented interface for users. | Which products, tasks, systems and users are covered by current availability and entitlements. |
| Joule Agents and Joule Studio | Support specialized task execution or coordination, and creation and management of custom skills and agents. | What the agent can actually do, which actions require approval and how exceptions are handled. |
| SAP BTP | Provide extension, integration and application-development capabilities. | Required services, integrations, skills and operating costs for the use case. |
This architecture is SAP’s stated direction, not proof that all components already work as one seamless layer in every customer environment. Availability and integration depend on the products, editions, configuration and contracts involved.
What Joule does—and what “everywhere” means
Joule is SAP’s user-facing assistant, not simply another name for an autonomous agent. SAP describes it as a way to interact with business information in natural language, get assistance inside applications and coordinate with specialized agents. Its broader aim is a more persistent experience across SAP tools and selected non-SAP contexts, including an action bar powered by WalkMe. SAP also described access through Microsoft Teams and Microsoft 365 Copilot in later-2025 updates.
“Everywhere” is best understood as a product direction, not universal coverage. Joule’s availability depends on the application, cloud edition, rollout, geography, customer configuration and entitlement. A customer should verify the exact supported tasks and systems rather than assume that one assistant can see or act across the whole enterprise.
Why Business Data Cloud and the Knowledge Graph matter
An AI assistant can produce a fluent answer and still misunderstand what a business term means. “Customer,” for example, may refer to a legal entity, a sold-to party, a service account or a group of related records. The answer becomes more useful when the system can connect current operational data to consistent definitions, relationships, lineage and access rules.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
SAP positioned Business Data Cloud as a foundation for trusted data products, analytics and intelligent applications, with access to SAP and external data. It said it would deliver hundreds of new data products during 2025 across areas including inventory, delivery, sustainability, compliance and workforce. SAP also announced People Intelligence based on workforce composition, skills and compensation data products from SuccessFactors. Delivery, availability and customer entitlements can differ by domain. SAP’s Business Data Cloud announcement describes the intended role of data products in intelligent applications.
The Knowledge Graph is intended to add semantic relationships. A search system might find documents mentioning a product and a customer. A modeled business graph could also represent that the product appears on an order, is associated with a delivery, and is linked to a service ticket for that customer. That context can help Joule retrieve more relevant information or support a more appropriate next step. SAP said its knowledge-graph capabilities could ingest and map metadata from SAP and external sources and consume models published in SAP Knowledge Graph on BTP.
Neither a data platform nor a graph repairs poor master data by itself. Duplicated records, conflicting definitions, missing integrations and incorrect permissions remain problems for the customer to address.
From an assistant to agents that move work forward
The important distinction is between talking about work and carrying it out. A Joule response might retrieve information or summarize a case; an agent could, depending on its design and permissions, coordinate steps such as collecting records, applying rules, requesting approval and initiating a transaction.
Recommended Free Tools
Rank #3
- Retrieval: find relevant business information.
- Summarization: assemble a concise view of a case or transaction.
- Recommendation: suggest an action based on available context.
- Workflow initiation: start a process or route work for approval.
- Transaction execution: make a change in a business system.
- Multi-step orchestration: coordinate work across applications, with controls for exceptions and approvals.
Those stages are not interchangeable. A demonstration of an agent recommending a pricing change does not establish that it can post the change in a live system, nor that it can do so without human approval. For finance, procurement, supply chain, HR or customer service, buyers should establish the agent’s permissions, approval thresholds, audit trail, escalation path and recovery procedure before production use.
Joule Studio: custom skills and agents
SAP described Joule Studio as part of SAP Build: a low-code/no-code environment to create, deploy, monitor and manage custom Joule skills and agents. The customer-build approach could let organizations tailor assistance to their processes, but it does not remove the need for developers, process owners and security teams to test what an agent can access and do.
At Sapphire, SAP targeted custom skills for the end of Q2 2025 and associated AI-agent capabilities with Q4 2025. These were announcement-era timelines, not a guarantee of access for every tenant on those dates. Check the current release status, supported scenarios, licensing and regional availability with SAP. SAP’s BTP and Joule Studio announcement provides the original capability and timing context.
What was announced, and what was available?
The Sapphire announcements covered different readiness levels. An existing feature, a planned release, a partnership and a demonstrated capability are not equivalent. The table reflects what SAP described around the 2025 event; it does not confirm present-day access in a particular customer’s tenant.
| Capability | Sapphire-era status | Qualification |
|---|---|---|
| Joule in selected SAP applications | Existing and expanding | Availability varied by application, edition, geography and entitlement. |
| Joule with SAP Business Data Cloud | SAP described it as generally available | Confirm the current product scope and applicable customer access. |
| People Intelligence | Announced as an intelligent-application capability | Announcement did not establish universal commercial availability. |
| Joule Studio custom skills | Targeted for end of Q2 2025 | Announcement target; not proof of access for every tenant. |
| Joule Studio AI agents | Associated with Q4 2025 availability | Verify current release, supported use cases and licensing. |
| Business Data Cloud on AWS | Announced from Q2 2025 | SAP expected Google Cloud and Microsoft Azure availability to follow in the second half of 2025. |
| Business Data Cloud data products | Hundreds announced for delivery during 2025 | Delivery and entitlement varied by domain. |
| Perplexity integration | Partnership and demonstrated capability | A demonstration or partnership announcement does not establish production access. |
| Palantir collaboration | Partnership announcement | SAP described collaboration around data harmonization, cloud migration and modernization; it is not evidence of a finished product. |
| Productivity gains of up to 30% | SAP assertion | Not an independent benchmark or guaranteed customer outcome. |
For a dated follow-up on SAP’s Business AI releases, see its Q2 2025 release highlights. Release summaries still do not replace checking the specific commercial terms and feature availability that apply to a customer.
Partnerships and the wider ecosystem
SAP’s partner announcements illustrate an ecosystem strategy, but each has to be judged by its actual delivery status. Perplexity was intended to help Joule work with unstructured information alongside SAP business data. Palantir’s announced collaboration focused on harmonization, migration and modernization scenarios. Deloitte planned an integration involving Zora AI and SAP Business AI. Later-2025 updates described access to Joule through Microsoft Teams and Microsoft 365 Copilot. SAP also announced Business Data Cloud availability on AWS, with Google Cloud and Microsoft Azure expected to follow in the second half of 2025.
These statements describe different kinds of commitments—partnerships, plans, demonstrations and availability announcements. They do not establish that every integration was generally available or included in an existing contract. SAP’s June 2025 partnerships and innovation update adds context; customers should confirm current product status before treating an announced integration as deployable.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.What SAP customers need before AI can help
Usable, governed data
Check whether customer, supplier, product, employee and finance records have owners, consistent definitions and traceable sources. Confirm that integrations are current and that the relevant data products exist for the business domain. Without this foundation, AI can make answers faster without making decisions better.
Free tools Windows power users keep installed
One-click scans. No signup required.
Best Value
Standardized processes and a realistic cloud fit
Embedded AI is likely to be easier to apply where cloud processes are relatively standard. Highly customized ECC or S/4HANA environments may need remediation, extensions or use-case-specific configuration. Customers substantially on-premises may find that access or integration depth differs from SAP’s cloud-focused direction. This does not mean migration is mandatory in every case, but an AI initiative alone is a weak reason to undertake a major ERP move without a wider business case.
Controls proportionate to an agent’s authority
- Role-based access that limits the data and transactions available to each agent.
- Human approvals for consequential or high-risk actions.
- Audit logs and clear ownership for decisions and errors.
- Escalation for ambiguous cases and a way to reverse actions where possible.
- Testing against missing data, conflicting records, unusual currencies, regional rules, outages and permission boundaries.
Clear commercial scope
Do not assume SAP software ownership automatically includes every Joule capability. Entitlements can depend on product, cloud edition, user type, message volume, AI units and the order form. SAP’s 2025 AI Services List referenced access to 2,500 Joule messages per Full Use in specified RISE Cloud Service contexts; that figure is not a universal allowance for SAP customers. Check the applicable SAP AI Services List and the customer’s own contract.
Does the strategy create a reason to buy more SAP cloud?
Potentially—but the platform decision should follow the business need. SAP’s advantage is its proximity to enterprise transactions and business semantics. A well-integrated agent operating on governed finance or supply-chain data may have a clearer path to changing a process than a generic assistant disconnected from those systems. The trade-off is greater dependence on SAP’s cloud, data platform, integrations, licensing and product roadmap.
The total cost is broader than an AI feature: cloud ERP or RISE, Business Data Cloud, BTP services, integration and data engineering, implementation, governance, change management and ongoing monitoring can all matter. Existing SAP customers with a cloud roadmap are the most natural fit for evaluating the strategy. Organizations seeking a vendor-neutral assistant, operating mostly outside SAP or unwilling to expand platform dependence should compare alternatives against their actual processes, not against keynote demonstrations.
How to evaluate a pilot
- Select one process. Favor high-volume, repetitive work with clear rules, useful data and measurable outcomes; keep humans in the loop for consequential decisions.
- Record a baseline. Measure cycle time, manual touches, error or exception rates, and the business outcome that matters before introducing AI.
- Map the data. List source systems, owners, definitions, data quality issues, permissions and required integrations.
- Define the agent’s authority. Separate what it may retrieve, recommend, initiate and execute; set approval points and escalation rules.
- Test difficult cases. Include missing fields, conflicting records, unusual currencies, compliance requirements, outages and unauthorized requests.
- Review results and controls. Compare against the baseline, inspect auditability and user adoption, and expand only if the benefit holds up under exceptions as well as routine cases.
Was Sapphire 2025 a real strategic shift?
Yes: SAP presented a durable direction toward applications that combine business data, semantic context and AI-assisted or agent-led work, rather than treating AI as a separate chat feature. SAP continued to describe a related autonomous-enterprise direction in 2026, a sign of strategic continuity—not evidence that later capabilities were available at Sapphire 2025. See SAP’s 2026 account of the autonomous enterprise.
The unresolved test is execution in real customer environments: whether data and processes are ready, whether agents can work safely across the landscape, whether access and costs are clear, and whether measured business results justify the platform investment. The headline is a strategy, not a production outcome.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
