SAP has introduced an Industry AI portfolio aimed at enterprise problems where better AI-generated recommendations alone are not enough.
SAP points to challenges such as coordinating thousands of field technicians after an energy-grid outage, keeping production lines running through supply-chain disruption, and automating pharmaceutical batch release without compromising quality or regulatory controls.
Dominik Metzger, president of Industry AI at SAP, put the challenge plainly. “These problems are incredibly hard to solve,” he said.
What Is SAP Industry AI?
SAP describes Industry AI as its approach to building AI around the processes, rules, data, and exceptions that define individual industries. Rather than starting with a general-purpose model and adapting it afterward, SAP wants industry context built into the applications and agents carrying out the work.
The ambition is to move beyond AI that summarizes information or recommends a next step. Industry AI is intended to support systems that can increasingly coordinate decisions and actions across complex business processes.
How Is SAP Industry AI Different From SAP Business AI?
SAP positions the Business AI Platform as the technical foundation beneath Industry AI. It provides the models, data, integration, and governance capabilities needed to build and run AI across the SAP landscape.
Industry AI adds the vertical specialization. Joule, meanwhile, serves as the engagement layer through which users interact with assistants and agents. In simple terms, the Business AI Platform provides the foundation, Joule provides the interface, and Industry AI supplies the industry context needed for specialized execution.
How Does SAP Industry AI Work?
SAP Business Data Cloud supplies contextual enterprise data, while SAP Domain Models draw on the company’s institutional knowledge to help AI systems understand business concepts and relationships.
SAP AI Agent Hub provides the management layer. Businesses can set boundaries around how agents, applications, large language models, and MCP servers operate and measure results against defined goals.
Industry AI brings those components into specific operational scenarios, where agents can use current business data and industry knowledge to work across processes that may span several applications and functions.
What Industry AI Applications Is SAP Building?
SAP has identified seven priority Industry AI domains: Asset Management, Commodity Management, Adaptive Production, Regulated Manufacturing, Revenue Growth Management, Unified Commerce, and Project Delivery.
The range shows that SAP is applying the model well beyond manufacturing. Asset Management targets reliability, uptime, safety, and compliance in asset-intensive businesses. Adaptive Production connects complex configure-to-order and engineer-to-order workflows. Unified Commerce brings together merchandising, planning, marketing, shopping, and fulfillment.
Each domain takes a business process that crosses traditional application boundaries and applies AI to the decisions and actions required to keep it moving.
Why Is SAP Using Forward-Deployed Engineers?
SAP’s forward-deployed engineering model embeds specialists, including data scientists and AI builders, directly with customers to tackle problems that do not yet have packaged answers.
SAP says it is beginning with select ECC and Private Cloud customers, using those engagements to test and refine solutions in real operating environments. Where an approach proves repeatable, SAP intends to turn it into standardized systems of agents that can be sold more broadly.
That makes forward-deployed engineering more than a services model. It is also part of how SAP plans to identify which Industry AI use cases are ready to become products.
How Does Industry AI Fit Into SAP’s Autonomous Enterprise Strategy?
SAP describes the Autonomous Enterprise as the broader destination: people set direction, assistants coordinate, and agents execute, with actions governed and measured against business outcomes.
Industry AI gives that model vertical depth. The idea is that an agent working in utilities, manufacturing, retail, or another sector should understand more than the generic mechanics of a workflow. It should also understand the industry rules and operating context that determine what a valid decision looks like.
That is the larger shift SAP is pursuing: from enterprise software that primarily records activity and supports decisions toward systems that can increasingly carry out parts of the work themselves.
What Should SAP Customers Watch as AI Moves Toward Autonomous Execution?
The first question is how much authority customers are prepared to give agents. Recommending an action, initiating a workflow, and independently executing a consequential business decision require very different levels of oversight.
Governance therefore becomes part of the buying decision. Customers will need to understand how agent permissions are defined, how actions are monitored and audited, when humans remain in the loop, and how an incorrect action can be stopped or reversed.
The other question is scalability. SAP’s forward-deployed model depends on turning customer-specific work into repeatable products. How quickly it can make that transition will help determine whether Industry AI becomes a broadly deployable software portfolio or remains concentrated in highly customized engagements.
What This Means for ERP Insiders
Decision authority is moving closer to agents. Customers should build testing, human oversight, escalation paths, and rollback controls into deployment plans as agents take on more operational responsibility.
Governance becomes a procurement checkpoint. IT, security, and risk teams should evaluate how agent permissions, actions, and audit trails are controlled before expanding autonomous execution.
Early Industry AI deployments may look different from conventional software rollouts. SAP is developing some of these capabilities directly with ECC and Private Cloud customers before standardizing them for wider use.
This article was originally published by SAPinsider.
