Skip to main content

Agents for Structured Data

Last updated on by vedant-aero-ml

Agents for Structured Data empower users to interact with extensive datasets through natural language queries. These applications seamlessly integrate vast amounts of data stored across heterogeneous enterprise systems, leveraging SAP Datasphere for federated data access, eliminating the need for data replication. By leveraging Agents, natural language inputs are first interpreted and then transformed into technical queries to retrieve pertinent data, with results delivered in an actionable format. The architecture also supports the streamlined integration RAG pipelines for query pre-processing and overall response optimization.

Agents in this scenario can cater to two core use case ideas: Descriptive and Prescriptive analytics. Descriptive analytics involves deriving insights and trends from data, while prescriptive analytics takes this a step further by offering proactive recommendations and actionable strategies based on the analyzed data, helping organizations optimize decision-making and operational efficiency.

Architecture​

image of solution diagram
Copy to clipboard

Solution Diagram Resources
You can download the Solution Diagram as a .drawio file for offline use. Alternatively, you may view and edit the Solution Diagram directly on draw.io.
Please note that any changes made online will need to be saved locally if you wish to keep them.
DownloadEdit

The architecture illustrates how Agents in this scenario operate within the SAP Business Technology Platform (BTP) ecosystem, enabling seamless integration of the AI component and structured data sources. At the core, the SAP Cloud Application Programming Model (CAP) serves as the orchestration layer, leveraging AI frameworks like LangChain and LangGraph to manage use case logic and data workflows in Code Based Agent approach. The agent can be designed in Project Agent Builder (PAB) and integrated with data in the Content Based Agent approach. More information about Code and Content based agents can be found here: Agent Streams.

SAP Datasphere plays a pivotal role by integrating with diverse data sources, federating data from SAP Cloud Solutions, third-party applications, or on-premise solutions. This allows agents to efficiently query and process large, distributed datasets without centralized storage. Meanwhile, the Vector Engine of SAP HANA Cloud supports a parallel RAG flow, enhancing search capabilities for real-time, contextually aware data retrieval, making the system well-suited for data-enriched enterprise applications. Data Federation ensures agility by enabling access to heterogeneous datasets without duplication, increasing efficiency.

These elements work together in unison to create an application that unifies data and AI, enabling real-time analytics and proactive decision-making. When combined with platforms like SAP Datasphere, agents can drive value across industries such as supply chain, logistics, financial services, and operations.

Services & Components​

For a comprehensive list of services, components and descriptions, please explore the Introduction on Services & Components.

  • SAP Datasphere enables a business data fabric architecture that uniquely harmonizes mission-critical data across the organization, unleashing business experts to make the most impactful decisions. It combines previously discrete capabilities into a unified service for data integration, cataloging, semantic modeling, data warehousing, and virtualizing workloads across SAP and non-SAP data.

Example Use Cases​

  • Finance KPI Exploration - Descriptive
    Finance agents empower sales & finance teams to “ask” for key performance indicators across massive, structured datasets without manual SQL or BI modeling.

  • Procurement Spend Classification - Descriptive
    Leveraging detailed purchase order and supplier master data, agents classify spend by category, vendor, and region—highlighting consolidation opportunities and non‑contracted purchases directly in a single NL query.

  • Replenishment Recommendation Engine - Prescriptive
    Agents combine inventory levels, sales forecasts, and lead‑time tables to calculate optimal reorder points and suggest purchase orders—automating replenishment planning to prevent stock‑outs.