Overview

SAP AI offers a comprehensive set of AI capabilities specifically engineered for integration into SAP's enterprise resource planning (ERP) and business application ecosystem. The platform targets developers and technical buyers within organizations that utilize SAP solutions, providing tools to embed artificial intelligence directly into core business processes. Key components include SAP AI Core, which serves as a runtime and lifecycle management platform for AI models, and SAP AI Launchpad, a central point for managing and monitoring AI scenarios.

The primary use cases for SAP AI involve automating repetitive tasks, enhancing predictive analytics within SAP applications, and facilitating data-driven decision-making. For instance, businesses can deploy AI models to optimize supply chain operations, improve customer service through intelligent chatbots, or automate financial forecasting. The system supports the deployment of custom machine learning models developed using various frameworks, allowing organizations to tailor AI solutions to specific business challenges while operating within a familiar SAP environment.

SAP AI is designed for enterprises seeking to operationalize AI across their business functions without needing to build extensive, separate AI infrastructure. Its value proposition centers on seamless integration with existing SAP data sources and applications, reducing the complexity often associated with AI adoption in large organizations. This integration capability allows developers to leverage established SAP security, data governance, and identity management frameworks, which is critical for maintaining compliance with regulations such as GDPR and adherence to standards like ISO 27001.

The platform provides SDKs for Python and Java, catering to common development environments for data scientists and software engineers. This extensibility allows for the development and deployment of a range of AI models, from traditional machine learning algorithms to deep learning networks. The focus remains on embedding these AI capabilities directly into the operational workflows managed by SAP systems, aiming to deliver measurable business outcomes.

Key features

  • SAP AI Core: A service for managing the lifecycle of AI models, including training, deployment, and inference, with support for various machine learning frameworks and runtimes (SAP AI Core documentation).
  • SAP AI Launchpad: A central application for monitoring, managing, and operating AI scenarios and models deployed on SAP AI Core, providing visibility into AI operations (SAP AI Launchpad documentation).
  • SAP Business AI: Pre-built AI capabilities embedded directly into SAP business applications, designed to automate specific tasks and enhance existing functionalities (e.g., intelligent invoice processing, predictive maintenance) (SAP Business AI product page).
  • Model Agnostic Deployment: Supports deployment of models developed with frameworks like TensorFlow, PyTorch, and Scikit-learn, allowing flexibility in model creation (SAP AI Core model deployment guide).
  • Integration with SAP Data: Direct connectivity to SAP data sources (e.g., SAP S/4HANA, SAP Data Warehouse Cloud) for training and inference, ensuring data consistency and security (SAP AI Core data integration documentation).
  • API and SDK Access: Provides REST APIs and SDKs (Python, Java) for programmatic interaction, enabling developers to integrate AI capabilities into custom applications (SAP AI API reference).
  • Compliance and Security: Built with enterprise security and compliance in mind, adhering to standards like GDPR and ISO 27001, and leveraging SAP's established security infrastructure.

Pricing

SAP AI Core offers a pay-as-you-go pricing model based on consumption. Costs are primarily determined by the usage of compute resources, storage, and data transfer. A free tier is available for initial development and testing.

SAP AI Core Pricing (as of 2026-05-28)
Resource Unit Cost Details
Compute (Training) vCPU-hour Billed per vCPU-hour consumed during model training jobs. Specific rates vary by instance type.
Compute (Inference) vCPU-hour Billed per vCPU-hour for model inference endpoints. Rates depend on the chosen inference instance.
Storage GB-month Billed for data stored in model repositories and associated artifacts.
Data Transfer GB Billed for data egress (outbound data transfer) from SAP AI Core.
Free Tier N/A Includes a limited amount of compute, storage, and data transfer for free to get started.

For detailed and up-to-date pricing information, refer to the official SAP AI pricing page.

Common integrations

Alternatives

  • Microsoft Azure AI: A suite of AI services and tools offering capabilities for machine learning, cognitive services, and AI infrastructure on Microsoft's cloud platform.
  • Google Cloud AI: Provides a range of AI and machine learning products, including Vertex AI for MLOps, pre-trained APIs, and custom model development tools.
  • AWS AI/ML: Amazon Web Services' portfolio of AI services, featuring Amazon SageMaker for machine learning development, as well as various pre-built AI services for specific tasks.

Getting started

The following Python example demonstrates how to interact with SAP AI Core using its SDK to list deployed scenarios. This assumes you have authenticated and configured your environment to connect to your SAP AI Core instance.

from ai_core_sdk.ai_core_v2_client import AICoreV2Client
from ai_core_sdk.models import ScenarioQuery, Scenario

# Replace with your SAP AI Core API URL and authentication token
AI_CORE_API_URL = "https://api.ai.sap.com/v2"
YOUR_AUTH_TOKEN = "YOUR_BEARER_TOKEN"

# Initialize the AI Core client
ai_core_client = AICoreV2Client(base_url=AI_CORE_API_URL, auth_token=YOUR_AUTH_TOKEN)

try:
    # List all scenarios
    scenarios_response: list[Scenario] = ai_core_client.scenarios.query_scenarios(query=ScenarioQuery())
    
    if scenarios_response:
        print("Successfully retrieved scenarios:")
        for scenario in scenarios_response:
            print(f"  - Name: {scenario.name}, ID: {scenario.id}")
    else:
        print("No scenarios found.")

except Exception as e:
    print(f"An error occurred: {e}")
    print("Please ensure your API URL and authentication token are correct and have the necessary permissions.")