Leveraging Artificial Intelligence-Solutions for Enhancing SAP S/

Improve every aspect of your business with AI-powered insights, recommendations, and automation. Take advantage of AI services embedded in your business applications and benefit quickly from intelligent technologies. Significantly reduce manual effort, increase efficiency, and reduce errors. Improve the user experience and increase customer satisfaction.

Customer

Planed Gen AI powered Service Agent Assistance in Dispute Management (Dispute Management, SAP Enterprise Service Management) and digital Agent for Customer Self Service (SAP S/4HANA Utilities)

Reduce cost of customer service
Reduce service response time.

SAP Business AI for Complex and Configurable Products Recommendations

	Challenge Companies selling complex and configurable products find it challenging to: Focus on the end-customers needs as much time needed to prepare technical product specifications Shorten the quotation process time Train and onboard their new sales workforce Scale business globally and open new markets due to the lack of local sales engineers Provide their end-customers with the truly e-commerce buying experience
	Solution Recommend optimal product and configuration based on end-customer needs without requiring a deep technical product ad market knowledge. Enable decision support by predicting commercial attributes to ensure the recommended solution meets customer's business objectives. Purpose built AI/ML framework with full lifecycle support: just load data, train model and publish, no need for a data scientist. Generative AI capabilities to perform AI-assisted customer requirements processing.
	Benefits Reduce time, effort & technical expertise required to generate a quote or a shopping cart Provide true e-commerce self-service by customer without the need for sales assistance Strengthen customer loyalty by recommending the best fitting solutions to the end-customer's needs Unleash the time of customer's sales experts to focus more on complex and high value opportunities Decrease time required for training sales reps and channel partners

Flowchart illustrating SAP Intelligent Product Recommendation (SAP IPR) leading to a sales quote document.

Business Value and Impact Overview Across Industries

Business Value	Business Impact Reference	Applicable Industries
Increase revenue by streamlining the quotation process Expand margins by insights from the recommendations Reimagine the customer experience by offering a 24x7 digital self-service assistants	Reduce manual effort for sales quote completion by 30-60 Reduce time to prepare a sales quote by 25-75	All industries dealing with complex and configurable products (Industrial Manufacturing, Mil industries, Life Science, Automotive, Wholesale, others)
**estimated value

Customer need-based Product Recommendations (SAP Intelligent Product Recommendations)

Reduce quote generative effort.
Reduce dependence on sales reps.

Meter-to-Cash

Validation of implausible meter reading results (SAP S/4HANA Utilities)

Better and faster resolution of implausible meter readings
Higher cash flow

Automatically Resolve Implausible Meter Readings Using AI

	Challenge The Metering Expert has to assure the quality of the metering data before they can be used for further processes. There is a significant effort for clarifying the implausible meter reading scenarios before they can be released or further action is required.
	Solution Based on historical and empiric data the Machine Learning for Utilities Metering supports in calculating a release confidence. This figure helps the Metering Expert to focus on the most relevant cases only as the application of machine learning enables to understand agents' past behavior. AI either automatically releases or suggests a release of all noncritical exceptions, for efficient reduction of manual work.
	Benefits Reduce manual work Improve efficiency and service quality Free resource capacity to focus on more strategic tasks

Automatically Resolve Implausible Meter Readings Using AI

	Challenge The Metering Expert has to assure the quality of the metering data before they can be used for further processes. There is a significant effort for clarifying the implausible meter reading scenarios before they can be released or further action is required.
	Solution Based on historical and empiric data the Machine Learning for Utilities Metering supports in calculating a release confidence. This figure helps the Metering Expert to focus on the most relevant cases only as the application of machine learning enables to understand agents' past behavior. AI either automatically releases or suggests a release of all noncritical exceptions, for efficient reduction of manual work.
	Benefits Reduce manual work Improve efficiency and service quality Free resource capacity to focus on more strategic tasks

Automatically Resolve Implausible Meter Readings Using AI

Challenge

The Metering Expert has to assure the quality of the metering data before they can be used for further processes.
There is a significant effort for clarifying the implausible meter reading scenarios before they can be released or further action is required.

Solution

Based on historical and empiric data the Machine Learning for Utilities Metering supports in calculating a release confidence.
This figure helps the Metering Expert to focus on the most relevant cases only as the application of machine learning enables to understand agents' past behavior.
AI either automatically releases or suggests a release of all noncritical exceptions, for efficient reduction of manual work.

Benefits

Reduce manual work
Improve efficiency and service quality
Free resource capacity to focus on more strategic tasks

Flowchart showing an ML Solution for Meter Reading leading to generating release confidence and releasing automatically.

Business Value and Impact Overview Across Industries

Business Value	Business Impact Reference	Applicable Industries
Increase operational efficiency by reduction of manual effort for validating unplausible meter readings by 90-95%. With 5% of unplausible meter readings, a Utility supplying 1 million customers, can save up to 1.500 person days and approx. 900 k EUR personal costs per billing cycle.	Leveraging SAP Utilities’ embedded machine learning, capabilities, Enedis (France) customized and trained a model on their historical data. This resulted in a highly accurate prediction of the meter reading plausibility, significantly reducing manual verification time and boosting efficiency for Enedis.	Integrated Utilities, Distribution System Operators and Meter Data Service Providers.

Business Value

Business Impact Reference

Applicable Industries

Increase operational efficiency by reduction of manual effort for validating unplausible meter readings by 90-95%.

With 5% of unplausible meter readings, a Utility supplying 1 million customers, can save up to 1.500 person days and approx. 900 k EUR personal costs per billing cycle.

Leveraging SAP Utilities’ embedded machine learning, capabilities, Enedis (France) customized and trained a model on their historical data. This resulted in a highly accurate prediction of the meter reading plausibility, significantly reducing manual verification time and boosting efficiency for Enedis.

Integrated Utilities, Distribution System Operators and Meter Data Service Providers.

Solution information:

Predictive scenario UTI_MR_IMPLAUSIBLE to use machine learning capabilities to process implausible meter reading results.
Report REML_FILL_EABLCONSHIST calculates the required consumption characteristics and stores them in table TEML_EABLCONSHST
This report is a prerequisite for training a predictive model.
Based on the entries saved in the table, a release confidence value is determined for specific meter reading documents.
Report REML_MR_IMPLSBL_RELPRED, which provides consumption characteristics for forecasting, and contains the algorithm, training dataset, and target variable that you can use to train the user-specific predictive model.
Forecast scenario and report are linked to the IS-U application.
Customizing Table TEML_IMPLMRPARAM
The results of report REML_MR_IMPLSBL_RELPRED (document number, release recommendation, and an indicator that specifies whether the documents were released automatically) are stored in table TEML_MR_IMPLSBL.
However, the report REML_MR_IMPLSBL_RELPRED only takes into account data records with a corresponding entry in the table TEML_EABLCONSHST.
A release confidence is calculated for these data records.
If The Release Automatically option is selected and a value is entered in the Release Threshold field, records whose release confidence is above or equal to this threshold can be released immediately.
If this is not the case, the release confidence is only saved in the table TEML_MR_IMPLSBL.
Implausible meter readings that cannot be released because, for example, B. The data record is locked are released in a subsequent run.
Note that there must be sufficient historical data to calculate the consumption characteristics.
If the table TEML_EABLCONSHST does not contain any entries, the calculation cannot be performed.
To access the predictive scenario and model, ISLM (Intelligent Scenario Lifecycle Management) is used.
SAP Fiori app Correction of Implausible Meter Reading Results (same as transaction EL27)

For further developments and the current status, see SAP Road Map Explorer.

To define which training data is relevant, you must define a time period. For example, a business requirement could be to train the model on receipts that are less than two years old. The same applies to the offset between the release of a document and possible follow-up actions for which the document must be processed further. This can be, for example, B. the expected period between the release of the document and a possible reversal at a later point in time.

The machine learning period for implausible meter readings is defined in customizing table TEML_IMPLMRPARAM. (Adjustment using transaction SM30)

Processing of out sorted billing documents (SAP S/4HANA Utilities).

Better and faster resolution of out sorted billing documents
Higher cash flow

Automatically Release Outsorted Utilities Billing Documents

	Challenge The utilities billing process is highly automated, nevertheless a major manual part is the processing of outsorted billing documents. With every periodic billing run, ~2-3% of the resulting do not pass simple checks, for example, check for maximum bill amount. These billing documents must be processed manually by agents.
	Solution Machine Learning for Utilities Billing utilizes previous decisions from the agent, where outsorted billing documents were resolved or reversed, including subsequent actions on the documents (such as reversal after release). A prediction or release confidence is provided to the user and can be used as additional information to release or reverse an outsorted billing document. Furthermore it is applied in combination with a selected threshold to automatically release an outsorted billing document without any further user interaction.
	Benefits Improve cash flow by increasing speed in resolving exceptions and in subsequent bill creation Increase quality in exception resolution Increase operational efficiency through the reduction of manual work for the Billing Specialist (Utilities)

Flowchart depicting the process of handling 30,000 billing documents outsourced. Machine learning is applied, releasing 20,000 documents with 10,000 remaining. Manual processing by agents resolves all outstanding documents.

Business Value and Impact Overview Across Industries

Business Value	Business Impact Reference	Applicable Industries
Reduce manual effort for the resolution of outsorted billing documents by ~ 60%.	No current reference. A utility creating 10 Million bills per year with 250.000 outsorted documents could save more than 1,5 Million EUR personal costs p.a.	Utilities industry in regulated in liberalized markets

Machine learning uses previous user decisions to release outsorted IS-U billing documents.

Solution information:

Predictive scenario UTI_BI_OUTSORTED to be able to use the machine learning functions for processing outsorted IS-U billing documents.
The report REML_BILLING_OUTSRTD_RELPRED calculates the required release confidence values and provides the optional automatic release of outsorted billing documents based on the release confidence for a defined threshold.
Storage of Results in Table TEML_BILLG_OUTSD
Outsorted accounts that cannot be released because, for example, B. The data record is locked are released in a subsequent run.
Note that there must be enough historical data to create the forecast.
If the table TEML_OUTBILPARAM does not contain any suitable entries, the calculation cannot be performed.
To access the predictive scenario and model, ISLM (Intelligent Scenario Lifecycle Management) is used.
SAP Fiori app Correction of Outsorted IS-U Billing Documents (like transaction EA05)

For further developments and the current status, see SAP Road Map Explorer.

To define which training data is relevant, you must define a period. For example, a business requirement could be to train the model on receipts that are less than two years old. The same applies to the offset between the release of a document and possible follow-up actions for which the document must be processed further. This can be, for example, B. the expected period between the release of the document and a possible reversal at a later point in time.

The machine learning period for implausible meter readings is defined in the Customizing table TEML_OUTBILPARAM. (Adjustment using transaction SM30)

Matching of incoming payments (SAP Cash Application, add-on for contract accounting)

Less manual work
Less days sales outstanding

Predicting Likelihood of Late Payment

	Challenge Many organizations have high rates of uncollected revenue and write-offs Organizations spend highly on the collection processes to collect late payments and debts Collections departments have limited visibility into the history, behavior, and risks of their customers, making it difficult to personalize services and optimize collections
	Solution SAP S/4HANA for behavioral insights: Provides standard intelligent scenarios and catalog content for the payment collection use case Provides explanations of contributing risk factors Provides 360-degree graphical timeline of all interactions with the customer Integrates directly to FI-CA Intelligent Collections
	Benefits Increased effectiveness of debt management and collection Better understanding of customer situation and more customers service approach Track and fine-tune collections success metrics

Flowchart detailing the process: invoicing, billing, and collections processes, aggregating and transforming data in SAP S/4HANA, applying a Behavioral Insights ML Model, visualizing customer journey and analyzing risk scores and reasons, and integrating into FICA Intelligent Collection to tailor collection strategies and steps.

Business Value and Impact Overview Across Industries

Business Value	Business Impact Reference	Applicable Industries
Enhance collections strategies. Resulting in improved revenue collection. Improve customer experience and engagement and ability to treat them with more empathy.	2-5% increased Identification of late payers 2-10% reduction in total outstanding debt 3-5% increased Revenue Recovery from Debtors "It is critical to use technology to make the entire compliance process far more efficient – simply scaling up what the US Internal Revenue Service does today will not produce the desired results" Testimony of Charles O. Rossotti, Former IRS Commissioner (1997-2002) before the Senate Committee on Finance Subcommittee on Taxation and IRS Oversight May 11, 2021	Subscription billing (BRIM) in Q2 2024 Tax agencies (TRM) Public sector (PSCD)

Business Value

Business Impact Reference

Applicable Industries

Enhance collections strategies. Resulting in improved revenue collection.

Improve customer experience and engagement and ability to treat them with more empathy.

2-5% increased Identification of late payers
2-10% reduction in total outstanding debt
3-5% increased Revenue Recovery from Debtors

"It is critical to use technology to make the entire compliance process far more efficient – simply scaling up what the US Internal Revenue Service does today will not produce the desired results"

Testimony of Charles O. Rossotti, Former IRS Commissioner (1997-2002) before the Senate Committee on Finance Subcommittee on Taxation and IRS Oversight May 11, 2021

Subscription billing (BRIM) in Q2 2024

Tax agencies (TRM)

Public sector (PSCD)

Solution information:

See SAP Notes: 2929951 - Important Notes for SAP Cash Application for 1809 Releases
Add-on in FI-CA
Clearing control must be configured for payment allocation.
In some cases, not everything can be stored in rules and then manual processing must be carried out (transaction FPCPL).
App for Data Exchange with Cash Application - Training of Machine Learning Model
This app schedules a job that is used to train the model.
Job: Mass Activity ML01 Training, Transaction FPML_CASHAPP_TRAIN
App for Data Exchange with Cash Application - Inference of Clarification Proposal
Mass Activity ML02 Inference
Transaction FPML_CASHAPP_INF
Tables DFKKZML_INF_EXTR and DFKKZML_INF_PROP
Mass Activity ML03 Training Master Data
Transaction FPML_CASHAPP_MASTER
Table DFKKZML_MD_EXTR
Central settings, such as the machine learning period, are made in table TFK006ML.

Note

Business Value	Applicable Industries
Reduced churn, increased customer satisfaction and brand loyalty Additional revenue by selling related products and services	Utilities (Energy Suppliers, Integrated Utilities)

Asset & Service

Optimization of maintenance by asset failure prediction (SAP Asset Performance Management)

higher utilization
less unplanned downtime

Condition-based maintenance using AI enabled visual inspection (SAP Asset Performance Management)

higher utilization
less unplanned downtime

Remark: Object part can be any spare/part of equipment where you identify the damage. If you record object part in notification then that can be help to analyze in future, for example, in a breakdown analysis.

For further developments and the current status, see SAP Road Map Explorer.

Dispatching and planning optimization of field service technicians (SAP Field Service Management)

higher billable hours
less cost field service

Note

Analytics

Faster insights into your data using natural language (SAP Analytic Cloud)

higher trustworthy retrieval
less effort to gain insights

For further developments and the current status, see SAP Road Map Explorer.

Further Development

Enhancements are always published in roadmap viewer.

SAP Road Map Explorer

Planned Future AI/GenAI Offerings

Next Generation customer interaction with Conversation-driven Self Service Agent for Utilities	Simplify customer service and reduce processing time with Next Best Action/Offer for Utilities Customers
Predict likelihood of late payments for utilities services based on behavioral insights	AI-enabled visual inspection for asset condition monitoring
Prediction of assignment duration – using AI to estimate the expected real duration of activities based on time efforts logged by technician	SAP Joule-based explanation of complex market communication documents

SAP provides new premium AI capabilities, most of which use generative AI technology.

These features are available to customers moving to the cloud using RISE Premium +, or for customers who procure stand-alone AI units (conditions apply, but may be detailed later).

All customers continue to benefit from our "Basic AI" as part of the cloud application subscription.

The customer can also benefit from our BTP AI services to build a custom solution today with their SAP BTP CPEA.

Base AI vs. Premium AI Capabilities

Base AI: Functionality is included at no additional cost in licensed SAP Customer Experience solutions.

Premium AI: Capabilities will incur an incremental cost that will be paid for through the purchase and consumption of SAP AI Units.

Examples of AI Capabilities In SAP Customer Experience Solutions*

	Base AI	Premium AI
AI Segmentation & Send Time Optimization in Marketing
Case Categorization
Product Recommendations
Deal Intelligence
Joule messages (up to annual allocation)
Joule messages (exceeding annual allocation)
All generative AI use cases Including: CX AI Toolkit Sales Cloud v2 and Service Cloud v2 GenAI capabilities Additional use cases planned in 2024 and beyond

* Functionality and use cases planned and are subject to change. Please see Roadmap Explorerfor latest updates.

Check out these extra links if you want to learn more or deepen your understanding. They're packed with further information to enrich your knowledge.

SAP Business AI | Artificial Intelligence | AI For Business

Note

Please note that the video does not have an audio track and only displays a screen recording of the system. The audio will be added in the next iteration.

	Challenge Understanding customers’ needs & wants is key to proactively delivering relevant & personalized customer experiences, resulting in increased overall customer satisfaction and brand loyalty.
	Solution Suggest next-best action/service based on Utilities data and customer behavior for example, based on the recent consumption trend, suggest to reduce consumption to avoid bill shock or suggest an energy audit to rate the energy/heat efficiency of the house to avoid a high heating bill in winter. Suggest best possible energy products/services/bundles based on Utilities data (for example consumption, billing history, meter type) and customer behavior such past participation in energy efficiency and demand response programs.
	Benefits Reduce churn Avoid or reduce collections/dunning overhead Sell energy related products and services

Leveraging Artificial Intelligence-Solutions for Enhancing SAP S/4HANA Utilities

Artificial Intelligence-Solutions SAP S/4HANA Utilities