Episode #193
Introduction
In episode 193 of our SAP on Azure video podcast we talk about the Azure Center for SAP Solutions and how it helps you from the Azure Portal to create and operate SAP systems. We talked about integrating other Azure solutions like Azure Monitor or the Health Check with your SAP system. One cool thing that is availalble under the hood are APIs which allows you to fetch all this information from your SAP system. And this can actually be quite powerful. Trond Stroemme took these APIs and built a really neat Copilot scenario around it. Using the Azure AI Studio you can now build a simple Chat that allows you to interact with the Azure Center for SAP Solutions.
Find all the links mentioned here: https://www.saponazurepodcast.de/episode193
Reach out to us for any feedback / questions:
- Robert Boban: https://www.linkedin.com/in/rboban/
- Goran Condric: https://www.linkedin.com/in/gorancondric/
- Holger Bruchelt: https://www.linkedin.com/in/holger-bruchelt/
#Microsoft #SAP #Azure #SAPonAzure #ACSS #AI
Summary created by AI
- Azure Center for SAP Solutions (ACSS) APIs:
Trond showed how to use the ACSS APIs to fetch information about SAP systems on Azure and build a chat bot with prompt-flow and GPT-4.
- ACSS APIs: Trond explained the different types of ACSS APIs, such as list by subscription, get virtual instance, get application server, etc. and how they can be used to query various properties of the SAP systems, such as location, environment, product, VM size, kernel patch level, etc.
- Python code: Trond showed the Python code that he used to call the ACSS APIs, using a service principal for authentication and a workload management client for accessing the ACSS resources. He also showed how to process the JSON response and extract the relevant information.
- API manager: Trond demonstrated how to use the API manager in the Azure portal to explore the ACSS APIs, see the HTTP requests and responses, and get code samples for different languages.
- Prompt-flow and GPT-4:
Holger explained how prompt-flow allows to create conversational AI scenarios with GPT-4 and how to test and deploy them on Azure ML Studio.
- Prompt-flow: Holger described prompt-flow as a series of steps that link to each other and can be used to create different types of AI scenarios, such as chat, text generation, image generation, etc. He also showed how to use the prompt-flow designer to create, edit, and test the steps interactively.
- GPT-4: Holger mentioned that the underlying model for the chat bot is GPT-4, which is a large-scale language model that can generate natural and coherent text based on a given prompt. He also said that GPT-4 provides the conversational skills and context awareness for the chat bot, without the need to predefine entities and intents.
- Azure ML Studio: Holger explained how to use Azure ML Studio to set up an Azure OpenAI service, deploy a GPT-4 model, and connect it to the prompt-flow chat bot. He also showed how to test the chat bot from the Azure ML Studio interface or from an endpoint URL.
- Chat with ACSS demo:
Trond and Holger tested the chat bot and asked various questions about the SAP systems, such as number, location, environment, operating system, kernel patch level, etc. The chat bot provided comprehensive and contextual answers based on the ACSS APIs.
- Chat bot questions and answers: Trond and Holger asked the chat bot several questions about the SAP systems, such as how many systems they have, what are the system IDs, what system has the largest DB server, what are the operating systems, what are the kernel patch levels, etc. The chat bot answered the questions by using the information from the ACSS APIs and providing additional details and comparisons when relevant.
- Chat bot context and prompt: Trond and Holger also demonstrated how the chat bot can remember the context of the conversation and refer back to previous answers or systems. They also explained how the chat bot uses a prompt to understand the user’s intention and how to engineer the prompt with some examples.
- GitHub repository: Trond shared his GitHub repository with the chat with ACSS prompt-flow and the instructions on how to use it. He also suggested how to extend the scenario with other Azure APIs and components.