Configuring SAP AI Core and SAP AI Launchpad

• Gain peace of mind with SAP managed deployments

  With SAP AI Core, customers gain peace of mind as they benefit from a complete service that packages all dependencies and hides the complexity of building your own training and serving productive environment. Instead, SAP AI Core exposes simple API endpoints to embed AI into your business applications.

• Quickly embed AI in SAP applications and business processes

  Customers can take advantage of the standardized integration into SAP applications to quickly build and integrate their AI use cases into their business applications.

• Strike the right balance between costs and performance

  On the one hand, customers benefit from accelerated performance with GPU support to run their most resource-hungry use cases at scale. On the other hand, customers run efficiently while keeping control of their costs by leveraging built-in autoscaling and scale-to-zero as well as by choosing from a broad range of storage, CPU, and GPU service plans.

  SAP AI Core uses Kubernetes clusters which provide proven features for running container-based applications for training and serving AI models. Such a cluster can provide different resources according to the current requirements. E.g., GPU nodes for heavy duty AI models or running complex ML AI pipelines that require different resources per step. The Kubernetes infrastructure is very fast and flexible (scale up / down the containers which run the AI code).

• Model once and serve multiple customers

  Customers benefit from simplified training and inferencing with reusable templates and can deliver model templates for their customers or internal teams, to train the model on their own data.

To realize those added values, SAP AI Core and SAP AI Launchpad offer enterprise-grade features to productize and operate your AI models. The main capabilities that SAP AI Core provides are the orchestration of AI workflows, such as model trainings and batch inference, as well as serving model inference, so that models can make predictions.

To ease the shipment of new AI scenarios, the service offers continuous delivery capabilities and ensures tenant isolation with multi-tenancy. To safeguard your costs and to scale on demand, you only pay for the resources you used but can tap into scalability and increased performance powered by GPUs.

All of the functionality comes out of the box and is managed for you while retaining openness to any AI framework so that you can ship your AI scenario easily.

To connect to your business applications as well as to operate your AI scenarios in AI Launchpad, SAP AI Core provides the standardized AI interface, AI API, which provides a common framework for consuming and operating your AI scenarios.

During development your AI scenarios, you are supported with development tooling, such as the SAP AI Core SDK, and full flexibility in the choice of storage for your data and models.

• Centralize AI lifecycle management

  With SAP AI Launchpad, customers can connect to all AI API-enabled services and streamline the AI lifecycle management of their AI scenarios SAP AI Launchpad. Benefit from uniform model status as well as uniform training and deployment, regardless of the underlying technology. As such, SAP AI Launchpad will not only increase transparency but will also enable to drive innovation by re-using AI content.

• Monitor and continuously improve model performance

  Customers benefit from simplified model retraining to continuously improve their model performance. They can also productize existing training models of supported AI runtimes or trigger jobs and deploy models directly using SAP AI Launchpad. The different resource groups (tenant) are displayed in SAP AI Launchpad.

The SAP tenant concept provides a separation of data and workspaces of customers in an SAP system.

As SAP AI Core is based on a Kubernetes cluster, the technical separation is achieved by using Kubernetes Resource Groups. You can create resource groups to isolate ML workloads, for example, to separate users/ tenants.

There are 2 use cases:

• An SAP LOB provides embedded AI for different customers: The LOB is administrating the BTP tenant for SAP AI Core. In order to make sure that customer data and AI assets like models and trainings workflows are safe/secured, each "subtenant" i.e. customer is managed in AI Core using a separate resource group.

• A customer is using SAP AI Core directly: As the customer is using his own BTP Organisation (i.e. Tenant), data and AI assets are safe and cannot be accessed by other customers. The Customer can still use resource groups to separate data and AI assets for different departments by assigning a resource group for each department of the company.

https://help.sap.com/docs/AI_CORE/2d6c5984063c40a59eda62f4a9135bee/8aae6cbe2c0e4290954b8f61b4b355b7.html?locale=en-US