Introduction
In the realm of business intelligence, the ability to quickly and effectively access data is paramount. Yet, for many organizations, the process of data retrieval remains mired in complexity, often requiring specialized knowledge of database languages and IT intervention. This creates a bottleneck that can impede the flow of information and the agility of decision-making. This extension prototype presents a transformative solution: a system that allows users to interact with their business data through natural language.
By simplifying access to SAP S/4HANA's rich data landscape, it's not only democratizing data but also enabling a more dynamic and informed business environment.
This blog demonstrates how you can build your own S/4HANA extensions using SAP BTP and Generative AI Hub with intelligent solutions: Simple and secure and for any language!
Prototype in action
This video is taken from the prototype application running on SAP BTP Kyma using the Generative AI Hub. It connects to a S/4HANA 2022 FPS02 sandbox system, hence the data we query here is past "Best Practice" data.
Since it accesses CDS View Entities, the data retrieved is real-time data. In our case I've added 2 CDS view entities into a schema which of course can be done as needed for your extension use case.
Architecture and technology
The prototype uses the ODBC for ABAP driver which is a magnificent tool to safely get data out of SAP S/4HANA. I recommend to read the blogs mentioned in the references section. In the backend there's a Python application running on SAP BTP Kyma secured by XSUAA and the SAP Approuter.
For the purpose of this blog I've written a plugin to SQLAlchemy to reflect the ODBC for ABAP driver specifics. If you are interested to use this library or collaborate on it, contact me.
The prototype setup uses GPT-4 as LLM but I was able to make a few queries work instantly with Codellama 13b. It will likely perform better with a larger Codellama and improved prompting.
Architecture outline for SQL generation through generative AI models
Conclusion
By harnessing the power of the SAP BTP Generative AI Hub, we have unlocked a realm where natural language bridges the gap between complex databases and the insightful queries of business users. This prototype demonstrates that through SAP BTP extensions you can enhance your S/4HANA system, both on-premise and in the cloud, too!
As we stand at the cusp of this new era of human/ machine interaction , it is imperative to recognize the transformative potential that such technologies hold.
Call to Action
The future of business intelligence is here, and it speaks your language. Experience the power of this approach firsthand and witness the impact it can have on your organization. Embrace the opportunity to democratize data access, accelerate insight discovery, and safeguard your sensitive information—all within the robust framework of SAP BTP services.
Take the first step towards a smarter, more agile approach to data analysis. Explore the SAP BTP Generative AI Hub and the myriad of services designed to build side-by-side extensions with SAP S/4HANA. Join the vanguard of businesses that are not just adapting to change but driving it. Visit our website, reach out to our team, and begin your journey with SAP BTP today. Your data is speaking; it's time to listen!
References
Using the ODBC driver for ABAP on Linux
Tutorial: Using ODBC driver for SQL Service
- SAP Managed Tags:
2 Comments
