Sponsored by: Accenture | Accelerating Financial Close with SAP Databricks

Summary

SAP Business Data Cloud (BDC) unifies SAP Analytics Cloud, SAP Data Sphere, and SAP Databricks under one umbrella, with SAP Databricks serving as the AI and ML engine that extends SAP's core data capabilities with data science workloads.
Accenture implemented SAP Databricks for its own financial close operations, reducing the close cycle from five days to three days using predictive insights, automated commentaries, and real-time monitoring built on curated SAP data products shared via zero-copy Delta Sharing.
The architecture eliminates the need to replicate SAP data by sharing curated data products directly into the Databricks Data and AI platform, with Genie enabling natural-language financial forecasting and predictions sent back to SAP via Reverse Delta Sharing.

Chapters

FAQs

What is SAP Business Data Cloud and how does Databricks fit into it?

SAP Business Data Cloud (BDC) is SAP's modern data estate solution that brings together SAP Analytics Cloud for analytics and planning, SAP Data Sphere for data integration and warehousing, and SAP Databricks as the embedded AI and ML engine. SAP Databricks extends BDC's capabilities with data science and ML workloads, allowing organizations to run advanced AI on curated SAP data products.

How does zero-copy data sharing work between SAP BDC and Databricks?

BDC delivers curated data products in areas like finance and supply chain, which can be shared with SAP Databricks or Enterprise Databricks using zero-copy data sharing—eliminating the traditional requirement to replicate SAP data and rebuild models. This means teams work on governed source data directly without the cost and complexity of maintaining separate copies.

What AI-powered financial use cases did Accenture implement on SAP Databricks?

Accenture implemented predictive insights, automated commentaries, real-time monitoring, and fully integrated treasury operations for its own financial close process. The implementation also includes a Genie space for natural-language financial forecasting and Reverse Delta Sharing to send model predictions back into SAP systems.

How did SAP Databricks help reduce Accenture's financial close cycle?

By combining the Databricks Data and AI platform's AI and ML capabilities with SAP's curated data products through zero-copy sharing, Accenture shifted from reactive reporting to proactive, insight-driven decisioning. According to this video, this transformation helped reduce the close cycle from a five-day to a three-day cycle.

Full transcript

[00:08] Welcome everyone and thank you for joining this session. I'm Amit Joshi from Accenture's data and AI practice. I'm Amit Joshi from Accenture's data and AI practice. Um I I have a lot of SAP experience and I'm
[00:22] part of the data bricks business group working on most of the modernization engagements. Um Today we are going to talk about um want to share how Accenture is using SAP data bricks uh to fundamentally transform the speed
[00:35] uh and intelligence of our own financial process. So, we'll we'll talk about SAP BDC briefly and get into SAP data bricks capabilities and how it can extend the
[00:47] SAP data or core functionality with its own AI ML um you know, powerhouse. So, >>
[00:59] our own enterprise implementation. Um and everything you see here today, uh it's inside of our own finance operations uh in a productive space. So, three things I want to take talk to you uh and uh want you to walk away with. First, that SAP business data
[01:12] cloud is and why data bricks is its AI ML engine. Uh second, the specific AI powered finance use case which Accenture implemented. And third, the architecture patterns uh that made it work uh at the
[01:25] enterprise scale. So, let's get started. What is SAP business data cloud? So, I'm going to briefly introduce uh it's a more complicated than it looks on this slide, but it's a very simpler version of uh SAP BDC.
[01:37] So, be a business data cloud or BDC uh as we can refer, um it's SAP's answer to the modern data estate. So, it brings together three core capabilities, um the analytics um and
[01:49] planning, uh data integration uh and data warehousing using SAP data sphere, uh analytics and planning using SAP analytics cloud. These were SAP's existing solutions, which are now under the umbrella of SAP BDC.
[02:03] Uh and the third thing, the key component here, you see highlighted, which is SAP Data Bricks. Uh and that is what extends SAP SAP's capabilities with data science and ML workloads.
[02:15] So, the the key architecture inside here is what you see here in the middle box. It's called as curated data products. Uh which is SAP's delivering a lot of the out of the box curated data products in
[02:27] most of the important areas like finance, supply chain. Um and then that data can be shared with SAP Data Bricks or Enterprise Data Bricks using something called as zero copy data sharing. It's a very key capability. So,
[02:40] traditionally, we had to replicate SAP data, uh very complicated process, rebuild all the models in Data Bricks to make sense out of it. Now, the data products, the curated data products from SAP brings a lot of semantics and
[02:52] context along. And you don't have to replicate that data physically into Data Bricks to do any machine learning or AI workloads. That is all done through zero copy data sharing. So, it's SAP Data
[03:04] Bricks is also built within BDC eco ecosystem itself. So, if you get a BDC tenant, you can provision these different artifacts. And the SAP Data Bricks workspace gets provisioned within SAP
[03:16] BDC ecosystem itself. And if you're using Enterprise Data Bricks, you can use something called as SAP BDC Connect, which is a connector to do zero copy data sharing with the Enterprise Data Bricks.
[03:28] Um
[03:41] you're going to bring that data over into SAP Data Bricks, source the data other other non-SAP data from other, you know, platforms like Azure, uh, GCP, whichever platform your Data Bricks is hosted on, and then you
[03:53] can do all SAP and non-SAP combined reporting either within SAP BDC or taking the SAP data out into Databricks ecosystem, uh, enterprise Databricks, and do all the advanced analytics there.
[04:07] Um, now how Accenture is utilizing this, right? So, for Accenture, SAP Databricks powers the next generation of AI-driven financial operations uh, for our enterprise. Uh, we are not just the number one
[04:19] partner for uh, SAP and Databricks, but we have our own implementation as customer zero. Um, and that's the use case we're going to talk about here. So, the benefit it provides uh, is
[04:32] we implemented SAP BDC uh, for simplification of a lot of SAPs uh, you know, uh, ecosystem products we had, and now introduce SAP Databricks within SAP BDC BDC ecosystem uh,
[04:45] to accelerate the financial close. So, how does that help? So, first of all, it's streamlining the data data consolidation. So, that helps reduce the manual efforts of gathering data across multiple source systems during the
[04:57] period close. That's going to give a lot of acceleration, and that's where a lot of time is spent to bring the data together uh, for the financial analyst to look at it. Uh, the reduced manual tasks, so our pre-close solution alone is reducing the financial close time by 40%. It used to
[05:11] run for 5 days, now we are able to run it within 3 days, uh, and that too we are only uh, we have taken our existing models from our Linux ecosystem, the Python models, and deployed on SAP Databricks. Just by
[05:23] deploy deploying them as is, we have seen this much uh, gain, and we are further working on optimization, and we are seeing it almost going down to up to day and a half financial close time period. That's a that's a great amount of gain uh, which gives time for the
[05:35] financial analyst to spend time on doing the actual analysis and you know, uh, look at the anomalies. Now, that also helps enhance the reporting and accuracy. So, now we have all the automated pipelines.
[05:47] Data is replicating into SAP BDC through replication flows. We have those data products which are very intuitive and out of the box from SAP with a lot of context. And data gets into SAP BDC within almost like no pretty much
[06:00] real-time or no latency. And all the AML models are continuously like scheduled and running. So, we get all the real-time financial intelligence and lot of reporting accuracy here. So, this is the most
[06:12] exciting area because our treasury forecaster and intelligent cash solution has freed up almost 20% of the idle cash which now can be used for our global operations. So, that's a great amount of again gain. That's the second metric we
[06:25] are tracking very closely. And underpinning all of this, we have like fully governed SAP aligned ecosystem that provides SAP and Unity Catalog governs the you know, all the data assets. So, it's
[06:38] enterprise-grade AI and not just a data science experiment. So, I'm going to quickly run through this slide. So, this is our kind of high-level analytics journey. So, as you can see here, up until FY23 we had multiple
[06:50] tools like BPC which is running consolidation, Hana Studio which is which was our kind of existing data lake, BW on Hana which was for the use for management reporting. So, it was solid but pretty
[07:04] fragmented ecosystem, right? So, multiple platforms, manual hand-offs of the data which used to take almost 5 to 7 days to get to financial close. By FY24, we started moving a lot of
[07:16] these workloads to SAP's business technology platform. SAP Analytics Cloud was introduced for planning and then BW Hana remained our backbone for data warehousing. FY25 is where the exactly was the
[07:29] inflection point, right? So, we introduced BTP Data Sphere which was SAP's answer to data warehousing and integration. And then BTP BW Bridge to bring a lot of BTP assets or BW assets into the BDC
[07:42] ecosystem. So, BTP was also introduced for the JNAI because at that time we did not have full-blown BDC and SAP Data Bricks. Um and then then by the time we got to FY26, this is
[07:55] what we are in in our present world today. As you see here, there's a lot of simplification. There are products only used for specific purposes like for consolidation which is group reporting which is SAP's out-of-the-box solution.
[08:08] Uh we have Joule for agenting build and BTP JNAI but again that focuses a lot on SAP's agenting AI. For anything to do with enterprise level AI, we have Business Data Cloud which is providing you the data products, the
[08:21] reporting, planning, analytics, and AI/ML and JNAI workloads. This is happening within the single box. So, now this is like fully governed ecosystem. As you can see the SAP Data Bricks is embedded within the BDC um
[08:34] tenant itself. >>
[08:46] Data Bricks. So, those capabilities really help uh unified AI and data platform gives a SQL notebooks, Unity Catalog, and ML Ops in a single environment. The only word of caution is SAP Data Bricks is not it's an OEM
[08:58] version of Data Bricks. So, it's not a full-blown advanced engineering capabilities. It is limited to what is needed for AI running AI/ML experiments. Um but you can always choose again to use SAP BDC connect to uh replicate this
[09:10] data into uh enterprise Data Bricks. Uh end-to-end automation of uh across finance workflows. So, all four of the production AI models I'm going to show you which is treasury forecaster, treasury optimizer, risk scoring, and pre-co pre-close commentaries. They run
[09:23] on fully automated schedule pipelines um and has a lot of control tower type monitoring uh built on top of it. There is human in the loop uh but we're going to talk about it briefly. Uh zero copy data sharing we already talked about and then enterprise grade
[09:36] security, and then seamless integration with SAP BDC and BTP ecosystem. I'm going to spend a little time here on the architecture and then we're going to see a brief demo. Um
[09:48] So, as you see here SAP Data Bricks is integrating with is is within that BDC box, right? So, this is a very important aspect of this architecture. Uh it's fully integrated as I mentioned when you provision BDC or Business Data
[10:01] Cloud you will be able to create something called as a uh uh a platform integration ecosystem, right? So, you can provision uh SAP Analytics Cloud, you can provision
[10:13] Data Sphere. You can provision uh SAP Data Bricks and they integrate with each other and that's where that zero copy data sharing integration is established. So, the data movement becomes very easy. Data integration becomes very easy. Data
[10:26] is available at your fingertips pretty near real time. Uh your as
[10:39] promotions, employee information, uh contracts. And S/4HANA provides us with the valuable information like invoices, uh journal entries, and things like that. All of this information uh is replicated using Delta Share, zero copy Delta Share
[10:52] into SAP Data Bricks. And then we have these four different models we just talked about uh helping with our financial close. Uh once these these are pretty scheduled to run in automated fashion. Um
[11:05] Something like pre-close I want to talk about a little little bit in detail. This is where uh it provides the commentaries, automated commentaries. We are we have agents running on top of SAP data. They generate commentaries and then that data is shared back with the
[11:17] financial analyst for their review. It is almost up to 80 85% accuracy at the moment. So, the financial analysts are able to now spend time rather than writing those commentaries, review the commentaries, fill the gaps, and take
[11:29] necessary actions. This is where the the value is uh of this architecture. Uh and then once this the data is uh the data products are created within
[11:41] Databricks are now also we are able to share them back into BDC. So, this is called as a bi-directional zero-copy delta sharing. You first get the data data from SAP into Databricks, do all the AI/ML capabilities, and now you're
[11:53] able to publish that data back into BDC catalog. Either you can use that within BDC for doing all your reporting and analytics or you can use within SAP Databricks itself and build your own sort of dashboards on top of it.
[12:06] Um the other important callout here is uh the Business Technology Platform uh is kind of integrating This is SAP's core platform for data integration. So, all
[12:18] of these capabilities are also tightly integrated with BDC or BTP ecosystem uh for any SAP-specific agentic workloads, right? So, if you had to receive any outputs from SAP agents, you're going to receive them back using SAP Business
[12:31] Technology Platform. So, the the bottom line is um Databricks powers the key uh AI-based financial operations here. Uh as we also heard a lot of exciting announcements, we have Genie code, um a lot of the agentic
[12:44] enhancements which uh Databricks is bringing to the table will be available out of the box within SAP Databricks. So, the AI AI agent gateway and you need catalog AI gateway, um agentic workflows, all of those
[12:57] capabilities are available within your enterprise SAP Databricks. Sorry. SAP Databricks ecosystem um and then that helps, you know, uh build this as an exten- extension to your SAP core data
[13:10] uh to run your AI/ML workloads. Um with that, I'm going to switch over to a demo. >> With zero-copy delta sharing, you avoid costly replication and egress fees while
[13:22] ensuring your data is always current. Let's take a quick look at how this works in action with SAP Business Data Cloud and Databricks. SAP data can now be securely shared in minutes with zero
[13:35] data movement, creating a trusted governed data foundation ready for AI, machine learning, and real business impact. To illustrate the power of this integration, consider this predictive scenario built using SAPS/4
[13:49] HANA sales orders data. We'll see how SAP sales order data can be seamlessly shared into Databricks, transformed in a notebook, used to train and visualize forecasting models, and then published
[14:01] back for consumption in SAP and other reporting tools. The core of this scenario is the sales order data product in SAP Business Data Cloud, which represents curated sales order data with
[14:14] business semantics coming from SAPS/4 HANA sharing uses Delta Sharing, which allows secure access to tables without copying or replicating the underlying data. So, SAP data stays in place while
[14:28] Databricks consumes it
[14:41] how governed productized SAP data can be opened to lakehouse
[14:54] built-in AI and AutoML, we can forecast product sales directly on live SAP data without building
[15:06] Databricks Assistant, which lowers the barrier to building data pipelines and models on SAP data. Instead of handwriting Spark SQL, the user invokes the assistant and provides a natural
[15:18] language prompt describing which columns to select from sales order and sales order item, how to join them, and which materials to filter on. The assistant generates Spark code to
[15:30] pull the required columns, for example, material, net amount,
[15:43] highlights how
[15:55] library to produce a 12-month forecast for both quantities and revenue. Visualizations
[16:08] bands for max and min forecast values. The consolidated data set is materialized as a Delta table. This table appears in the Databricks catalog and can then be exposed as a data product and published back into SAP
[16:21] Business Data Cloud, making forecast results consumable by SAP Datasphere and SAP Analytics Cloud. A crucial capability showcased in this scenario is the ability to operationalize
[16:34] predictions and make them available back in SAP landscapes. Next step is to create a Delta share based on the sales_forecast_dp
[16:46] table. Then we use this Delta share and the BDC recipient to share this data product from SAP Databricks to SAP
[16:58] the product sales forecasted asset has automatically appeared. It inherits the schema and lineage directly from Databricks, making it immediately available for our finance and supply chain teams to consume in BDC. This
[17:12] reverse sharing demonstrates how Databricks
[17:28] Databricks. By partnering with Accenture, you gain a future-proof architecture that accelerates innovation and maximizes ROI across your enterprise. Ready to transform your SAP data into a
[17:41] strategic advantage? Contact us today. >> Um so that that's kind of how it works. Um and as you as you saw like we this is
[17:54] a little bit data demo. So you saw like AI system which is now Genie code. All the capabilities which we have seen Databricks has launched last in last couple of days in keynote will be available within SAP Databricks ecosystem. Or you can replicate SAP data
[18:07] using zero copy data sharing and utilize all the capabilities of the enterprise Databricks platform. Um that help That concludes the session. Um we have a booth right behind us of Accenture. If you want additional
[18:20] demos, deep dives into SAP BDC or Databricks integration, feel free to reach out to us. Thank you.

Learn more about the Databricks Data and AI platform.

The information provided herein is for general informational purposes only and may not reflect the most current product capabilities or configurations.