Three Steps To Orchestrating AI-Driven Process Transformation

AI adoption in the SAP ecosystem remains in its early stages, but the reason pilots stall is becoming clearer. In SAPinsider’s June 2026 Spotlight report, Orchestrating AI-Driven Process Transformation in SAP-Centric Enterprises, nearly three-quarters of respondents report that AI-enabled SAP use cases in their organizations are still in the identification, experimentation, or no-plans phase. However, the same research finds that specific categories are already delivering measurable value. Moreover, the decisive variable separating deployed applications from stalled demos is the execution layer that turns AI outputs into governed, deployed business applications running against SAP data.

That framing makes low-code/no-code platforms key in conversations around migration. In the survey, 39% of respondents describe these platforms as core or important to turning AI outputs into business applications. Neptune Software, the report’s featured sponsor, has spent the past year building its Neptune DXP platform to move from AI-assisted app generation to full AI orchestration and agent management within the SAP stack.

The report recommends three actions that organizations must take to successfully orchestrate AI processes during their digital transformation process:

Step 1: Start Where Value Is Already Proven

The report’s first required action is to anchor the AI strategy in value-proven use-case categories rather than treating adoption as a discovery exercise. Workflow automation and task routing (30%), conversational interfaces (28%), decision support (24%), content generation (22%), and predictive analytics (22%) lead the list of use cases delivering value today.

These categories require AI output to become a working application embedded in a real process, a gap that Neptune’s Naia Build addresses.

Step 2: Authorization Alignment Is the Test

The report is unambiguous about what gates the path to production. Two-thirds of respondents (63%) cite accuracy and reliability of AI outputs in critical processes as their top concern, with data leakage through AI services (59%) and data privacy and regulatory compliance (59%) close behind. When asked about security controls for AI using SAP data, 49% of respondents cited role-based access control aligned with SAP authorizations as the single most important requirement.

These findings highlight that approaches which replicate SAP data into external stores must rebuild governance around the copy. They must add latency, cost, and a new surface to secure. For example, applications and agents built on Neptune DXP run within SAP, using the customer’s own data and business logic, inheriting the system of record’s existing role and authorization model.

The governance question extends beyond apps to agents. Thus, for enterprise architects weighing the report’s second required action of implementing governance controls before any production scale-up, Neptune DXP’s observability and auditability capabilities determine whether an agent acting on operational ERP data can be reviewed, constrained, and trusted.

Step 3: The Execution Layer as an Architectural Decision

The report’s third required action asks SAPinsiders to treat the execution layer as their most important architectural decision and to evaluate candidates against three criteria:

• SAP-aligned role-based access control at the application level
• Fit with dominant integration patterns.
• Observability against the metrics that organizations track (employee productivity at 38%, customer outcomes at 28%, and process KPIs at 28% per the SAPinsider report)

In terms of integration, the survey found that direct API calls (33%) and SAP BTP AI Core or AI Foundation (22%) are the most common ways to connect AI services to SAP backends. The Neptune DXP 24.14 release maps to this pattern with an Integration Hub for SAP solutions, extended SAP BTP Proxy capabilities, and a secure Vault service for credentials and secrets. These features position the platform to orchestrate AI across SAP and non-SAP systems such as Salesforce, Microsoft, and ServiceNow without moving governance outside the SAP landscape.

What This Means for SAPinsiders

The execution layer decides whether AI pilots reach production. The SAPinsider report shows value concentrated in use cases deployed within real processes, while 43% of organizations are not yet using AI in SAP-related processes at all. Enterprise architects and ERP program managers should now evaluate execution-layer candidates, including Neptune DXP, against the report’s criteria: SAP-aligned access control, integration fit, and observability. Selecting this layer before scaling pilots separates funded roadmaps from stalled proof-of-concept portfolios.

Inherited governance beats bolted-on governance. With 63% of respondents concerned about accuracy and 59% about data leakage, the standard for systems of record is full reliability under existing controls. Platforms like Neptune DXP that run within SAP and support the native authorization model let governance teams extend proven SAP role concepts to AI, rather than rebuilding compliance around replicated data. GRC leaders should make SAP-aligned role-based access control the top-cited requirement per SAPinsider and a non-negotiable evaluation criterion.

Focus funding on value-proven use case categories and instrument them from day one. AI is expensive to run, and the organizations showing value first are those funding two or three proven categories, such as workflow automation, conversational interfaces, or decision support, mapped to existing SAP processes. Development leaders should pair each funded use case with the measurement approach the market already uses, such as productivity gains, customer outcomes, or process KPIs, and use capabilities like Agent Trace to make those results visible to the business from the first deployment.