ad
ad

? How to build a GenAI application using SAP BTP, Generative AI Hub and Amazon Bedrock

Science & Technology

Introduction

Welcome to the second week of the SAP PHOB Fest! In this session, we're diving into how to build a generative AI application using SAP BTP, specifically focusing on the Generative AI Hub and Amazon Bedrock.

Event Overview

Before we proceed, be sure to check out the SAP Community page for various activities throughout our four-week event. Participating can earn you points toward the grand prize: a retreat at one of SAP’s Road Show stops in 2025!

Introduction to Today's Speakers

We're pleased to have Abby and Raj, joining us from Amazon Web Services (AWS). They will walk us through our partnership and showcase which models can be accessed via the Generative AI Hub in SAP’s Business Technology Platform (SAP BTP).

Agenda

Today's session will feature two parts:

  1. Understanding Why: Why should you consider building your generative AI applications with Bedrock models and the Generative AI Hub?
  2. How-To Walkthrough: A comprehensive guide on how to set up the Generative AI Hub to consume Bedrock models within your applications.

Throughout the presentation, we'll present several QR codes. Feel free to pause the video to capture these links to GitHub resources and documentation.

Why Choose SAP's Generative AI Hub?

SAP's Generative AI Hub is a centralized platform that offers carefully selected models from which you can effectively build applications. These models have been vetted with SAP-specific guardrails and orchestration capabilities, ensuring that companies can jumpstart their business use cases while focusing on building rather than baseline compliance aspects.

You can thrive in a compliant approach to access various large language models (LLMs) under this SAP ecosystem, helping enable advanced use cases such as retrieval-augmented generation (RAG) and agentic workflows.

Amazon Bedrock's Contribution

Amazon is integral to this ecosystem, providing powerful models via Amazon Bedrock, including the Titan and Anthropic Claude families. These models deliver exceptional performance and are frequently recognized on industry leaderboards. Specific models available through the Generative AI Hub include Claude 3, Claude 3.5, Sonnet, and Titan.

You can track updates about available models through the Generative AI Hub documentation.

Exploring Model Availability

The Generative AI Hub is available in several AWS regions, including Sydney, Frankfurt, Tokyo, and Virginia. Customers can benefit from numerous model options, including cost-effective alternatives compared to competitors.

For instance:

  • Claude 3.5 is about 3% cheaper than GPT-4.
  • Claude 3 Haiku is approximately 48% cheaper than GPT-3.5 Turbo.
  • Titan Text Light is 66% cheaper than GPT-3.5 Turbo.

Performance and Safety Features

When using Amazon’s models, keep in mind their emphasis on safety, responsible AI, and explainability. Many customers utilize the Claude family as front-end moderators, handling content moderation and compliance checks before more specialized models are employed.

How to Set Up Generative AI Hub and AI Core

To get started, follow these key steps:

  1. Verify SAP BTP Entitlements: Check for the appropriate service plans in your subaccount.
  2. Set Up Roles and Spaces: Ensure you have created a Cloud Foundry space and assigned appropriate roles for efficient access.
  3. Use Boosters for Configuration: Access the SAP AI Core and AI Launchpad via boosters to streamline setup procedures.
  4. Create Service Keys: These keys facilitate access to AI Core models.
  5. Deploy Models: Create configurations and deployments for model access.

Consuming Bedrock Models

To consume models, users can interact with various APIs provided by AI Core. The session demonstrated how to perform simple requests with the Anthropic and Titan models, showcasing capabilities like generating SQL queries from natural language.

Using the Generative AI Hub SDK, users can simplify interactions with these models, enabling quick and effective implementations of generative capabilities.

Conclusion

As you embark on your journey with generative AI and SAP BTP, consider these insights and practices. Testing various models and deployment scenarios can help you find the best fit for your business needs.

Thank you for attending today’s session. We hoping you learned something valuable and we encourage you to explore these offerings and provide feedback.

Keyword

Keywords: Generative AI, SAP BTP, Amazon Bedrock, AI Core, Anthropic models, Titan models, cloud computing, use case, deployment, compliance.

FAQ

Q: What is the Generative AI Hub?
A: The Generative AI Hub is a centralized platform within SAP BTP for accessing vetted models and building generative AI applications.

Q: How does Amazon Bedrock contribute to SAP’s offerings?
A: Amazon Bedrock provides a suite of powerful models, including Titan and Anthropic families, that are essential for building AI applications within SAP environments.

Q: What areas can generative AI models be employed?
A: Generative AI can be applied in various industries, including healthcare, finance, and advertising, to enhance efficiency and creativity.

Q: How can I ensure compliance while using AI models in my business?
A: SAP's Generative AI Hub includes built-in guardrails and compliance checks, making it easier to utilize models while adhering to legal and business requirements.