Article

Enabling Artificial Intelligence in SAP

A framework for shaping your AI journey in SAP
Published

6 June 2025

Artificial Intelligence (AI) is improving daily and offers countless opportunities for organisations to add value to their businesses. Enterprise Resource Planning (ERP) systems are in an interesting position to drive AI enablement as they contain large volumes of data and support daily operations. The biggest player in the ERP market is SAP, who has introduced an AI-First strategy along with SAP Business AI. Effective AI integration in SAP requires a clear strategy, leveraging embedded features, and a bottom-up approach to meet actual business needs. This involves mapping business processes to identify pain points, evaluating AI use-cases based on feasibility, and prioritising use cases.



AI in SAP


According to a Gartner study from 2024, 74% of CEOs believe AI will significantly impact their industry. However, nearly half of CIOs report that AI has not met the expected ROI. The gap in expectations and outcomes only serves to highlight the complexity of AI implementation and the need for businesses to navigate the AI hype. 


The first step is understanding what is behind the term Artificial Intelligence. The figure below illustrates an overview of AI and its various subcategories.

ERP systems are ideal to enable AI for their customers as they contain large quantities of data and drive daily operations. SAP recently announced SAP Business AI which aims to embed AI solutions across their cloud product portfolio, support users with their AI co-pilot Joule, and enable customers to create their own AI applications through the AI Foundation. SAP’s end goal is to apply generative AI to every task. 


SAP embedded AI is directly integrated into the SAP standard business applications, as out-of-the-box functionality and easy deployment. The figure below shows examples of embedded AI features. Underlying models used in SAP’s embedded AI solutions are available for reuse in the SAP AI Foundation allowing for the development of custom AI solutions.

Rather than adopting a top-down approach to shoehorn AI into a business, AI should be implemented from the bottom up – to ensure it addresses actual business needs. We suggest a three-step approach to break down the problem and find AI use cases that align with your business objectives. By focusing on a structured, yet flexible, and practical methodology, you can ensure that your AI initiatives drive tangible improvements and support your overall strategic goals.


Introducing the framework


The purpose of the framework is to enable organisations to identify and prioritise feasible AI use-cases within their SAP system. Here is a brief overview:

The method ensures AI initiatives are strategically focused, aligned with business needs, and maximises their benefits.


Step 1: Discovery 


To successfully leverage AI, organisations need to focus on use cases that offer the most value and impact. Start with a clear hypothesis about which business processes stand to benefit most – this helps narrow the search and drive meaningful results faster. Methods like SCOR or SIPOC can be utilised to get a thorough understanding of your business process. In addition, intelligent process management tools like SAP Signavio can accelerate the discovery of the AS-IS process landscape, uncover bottlenecks, and pinpoint areas for AI-driven optimisation.

The business processes must be mapped to a detailed level (level 3 or 4) for two key reasons:

  1. Identifying specific pain points
  2. Narrowing down data sources

Mapping business processes to a detailed level helps with clearly formulating the problems and specifying requirements for the solution. AI is heavily dependent on your data quality, therefore, narrowing down the data sources helps identify if potential data cleansing and standardisation is required. The result of this step is a whole list of use cases where AI may address pain points for the business.


Step 2: Evaluation


The next step involves a detailed evaluation of each use case or process step to determine if it is fit for AI and define the appropriate AI archetype. The evaluation considers two dimensions: complexity and repetition.


Complexity: Determine whether the process step is creative and nuanced or predictable and rule based. Creative and nuanced tasks often require a high degree of human judgement and innovation, making them more challenging to automate fully. These tasks might benefit from AI that augments human capabilities, such as decision support systems or advanced analytics tools. On the other hand, predictable and rule-based tasks are ideal candidates for automation. These tasks follow a set pattern or logic, making them easier for AI to handle with minimal human intervention.


Repetition: Evaluate the frequency and volume of the process step. High-frequency, repetitive tasks are prime targets for AI automation. Automating these tasks can significantly enhance efficiency by reducing time and effort. AI can handle these tasks consistently and accurately, allowing employees to focus on more strategic and creative activities. Conversely, tasks that occur infrequently or involve unique, one-off situations may be less suited for AI automation and require a different approach.


The figure below illustrates the categorisation of AI use cases by complexity and repetition.

AI execution enables full automation of tasks where AI can operate independently, while AI support enhances human decision-making through predictive analytics and recommendations. Robotic Process Automation (RPA) is ideal for automating repetitive, rule-based tasks, improving efficiency, and reducing manual effort. In cases where infrequent tasks require high-level judgement, creativity, or complex decision-making, a manual approach remains necessary.


Assessing each process step through the lenses of complexity and repetition allows organisations to define the most suitable AI archetype for each task. The targeted approach ensures that AI implementations are not only technically feasible but also strategically aligned with the overall business goals. 


Step 3: Prioritisation


Prioritising AI implementations involves assessing each potential AI use-case based on feasibility and impact.


Feasibility: Determine how easy or difficult it is to implement an AI use case. It is about the practicality of the implementation, and depends on factors such as:

  • Integration and development: Embedded solutions offer seamless integration and quick implementation but have limitations when it comes to customisation and may not meet business requirements. Customised solutions are tailored to meet specific business needs, providing greater flexibility but requiring more effort and resources to implement and maintain.
  • Data quality: Data must be available and of good quality for AI to produce reliable and accurate results. A data quality assessment is essential to understand if data cleansing and standardisation are required.
  • Change management: AI can significantly alter workflows requiring new skills for end users. The organisation must consider the extent of the training and change management investment required.



Impact: Determine the potential value or benefit of the AI use case to the organisation or end user. It basically starts with asking the question: "Is this worth doing?" 


The answer will depend on factors such as:

  • Profit generation and cost savings: The expected monetary improvements to the business such as sales increase through upselling or improved efficiency through automated tasks. Organisations must also weigh the benefits against the implementation costs such as license, development, training etc.
  • Scalability: If the solution can be rolled out across products or geographies.
  • Synergy: Another consideration is if the AI use cases share synergies improving the overall impact i.e. 1 +1 = 3.



The figure below illustrates the categorisation of AI use cases on feasibility and impact

The goal is to focus on Easy wins meaning high-impact, manageable projects to ensure a successful AI enablement journey and build AI capabilities. Subsequently, organisations may proceed with Quality of life use cases to further build AI capabilities or Strategic bets if resources are available and the benefits are high. If only last in line use cases remain, the organisation should consider repeating the Discovery and Evaluation steps for other business process areas to explore others with higher priority.


Prioritising AI use cases based on feasibility and impact allows organisations to focus their resources on building AI capabilities and delivering high impact quickly. In addition, it requires the organisation to consider the business case for each initiative to ensure that they meet actual business needs.


Final word


Successfully leveraging AI within SAP involves a structured approach encompassing the discovery, evaluation, and prioritisation of AI use cases. By starting with a thorough understanding of business processes and identifying specific pain points, organisations can pinpoint where AI can drive significant improvements. Assessing each step of the process through the dimensions of complexity and repetition ensures that AI implementations are feasible and support business goals.


When integrating AI solutions, the choice between embedded and customised solutions is crucial. Embedded solutions offer seamless functionality and easier updates at a lower cost. On the other hand, customised solutions provide greater flexibility to meet specific organisational needs despite requiring more effort to implement and maintain. Prioritising AI implementations based on their potential impact and feasibility helps organisations focus on high-impact, manageable use cases, ensuring a successful AI enablement journey.


Reach out if you would like feedback on your approach or input on how to initiate your AI enablement in SAP.

Related0 4