What SAP’s Autonomous Enterprise Demands from Sales

SAP’s Autonomous Enterprise vision sets a clear expectation: AI embedded in business operations should execute. SAP’s Autonomous Suite  deploys more than 50 Joule AI assistants across finance, supply chain, procurement, HR, and customer engagement, orchestrating more than 200 specialized agents. Sales workflows will follow.

Agentic AI makes that possible by acting without waiting to be asked. Rather than presenting data for a rep to interpret, an agent perceives a condition, reasons through a response, and executes a workflow on its own. The practical question is whether agents can be trusted to act on real customer data — without surfacing the wrong account, triggering the wrong promotion, or running outside the guardrails a business requires.

From Competition to Real-World Deployment

DataXstream, an Endorsed App Partner, placed 2nd among 13 global finalists at SAP’s Hack2Build competition in April 2026. The entry was the Churn Risk Agent, an automated workflow that identifies at-risk customers and triggers a retention response. The finalists included global consultancies Capgemini and accenture.

“Through Hack2Build, we leveraged SAP Joule Studio to quickly create an agent, integrate it with an MCP server, and deploy a suite of tools that enabled churn recognition and retention workflows grounded in system data and contextual business logic,” said John Kane, Senior AI Engineer at DataXstream.

SAP subsequently selected DataXstream for the SAP Agent Race to Sapphire, a program inviting a limited group of partners to showcase AI innovation at SAP Sapphire 2026. The challenge required DataXstream’s OMS+ IA team to design and deliver more than 20 intelligent agents — each covering multiple tool sets, with no functional overlap.

Inside the Churn Risk Agent

The Churn Risk Agent is a coordinated suite of agents built on SAP BTP, deployed via Kyma runtime with containerized machine learning services.

The workflow runs through SAP Joule Studio, beginning with the Churn Risk Tool, a proprietary ML model that evaluates each customer’s order history and activity patterns to calculate a churn risk percentage and rank accounts by priority.

Megan Herring, a developer at DataXstream, notes that the model can evaluate whether a 20-day gap between orders represents normal buying behavior for that specific customer or a warning sign — not a fixed threshold applied across all accounts.

Once the model flags an account, the Customer/Material Recommendation Agent analyzes that customer’s full buying history to identify which materials are most likely to re-engage them. The Material Availability Agent then checks current inventory and availability for those materials in real time. The Customer Pricing and Promotion Tool retrieves open promotions relevant to those materials and that specific customer, offering discounts that match the account’s purchasing profile and existing relationship with the business.

The output layer pulls all of that together. It creates an inquiry transaction directly in SAP with the customer and materials pre-populated, and drafts a retention email for the responsible sales agent. The agents have already evaluated the customer, identified products, confirmed availability, and surfaced a relevant promotion. The only human action required is a review and a click to send.

The Platform and Governance Behind the Agents

SAP Joule Studio has evolved from a basic chatbot with limited SAP connections into a multi-tier development environment that now supports external tool servers and Agent-to-Agent (A2A) communication. Hack2Build gave DataXstream access to functionality not yet publicly available at the time, accelerating what the team could build and test.

A2A communication is best understood as a smart middleware server. Traditional APIs fail when a single field is malformed or an expected value is missing; the A2A server understands context and resolves requests intelligently. That capability allows the agents to operate as a coherent, sequential workflow.

SAP’s governance layer sits underneath all of it. Kane describes SAP’s monitoring tools as the most granular in the industry, giving organizations visibility into which agents are accessing which data. Back-end connections are limited to pre-certified, safety-reviewed paths, a requirement for enterprises managing customer and pricing data inside S/4HANA.

Herring adds that SAP has prioritized both low-code and high-code paths for building agent skills in Joule. Organizations without deep developer resources can build and deploy agents using SAP’s guided tooling — not just those with large engineering teams.

“This achievement demonstrates how DataXstream is advancing OMS+ beyond traditional order management into a true sales execution platform powered by AI. By embedding intelligent agents directly into SAP workflows, we’re helping our customers turn data into action faster, more consistently, and with measurable business impact,” said Bryan Cain, Director of Artificial Intelligence at DataXstream.

What This Means for SAPinsiders

• Churn Risk Agent is available on the SAP Store. The agent is a SAP-validated implementation that organizations can deploy without building from scratch. Organizations actively exploring agentic sales workflows have a production-ready reference point to assess against their own SAP environments.
• SAP’s low-code Joule tooling lowers the barrier to agent development. Organizations without large dev teams are beginning to build their own agent libraries using SAP’s guided tooling. That accessibility is expanding who can participate in the Autonomous Enterprise build — not just system integrators and hyperscalers.
• MCP/A2A architecture is front-end agnostic, pointing toward a composable enterprise AI layer. Agent tooling built for OMS+ can connect to Joule Studio or a customer’s own chatbot, decoupling agents from any single interface. Enterprises evaluating long-term AI infrastructure are beginning to factor composability into platform decisions.