Enabling AI Transformation: Embedding AI into Business Processes

Article by Jonathan R. & Rohit Chandrasekhar

Sapphire has just passed. Everyone is talking about AI, including SAP themselves. Organisations have moved beyond “should we adopt AI” to “how should we embed AI into processes” and “how should we unlock the benefits from our AI investment”.?

Personally, I found the event was great, as I was able? to connect in person with clients and friends from across the SAP Industry ecosystem. My insights from SAP Sapphire 2024 emphasise the urgent pivot to AI for SAP. The focus on AI and the SAP Business Technology Platform (BTP) highlights how these technologies enhance efficiency and innovation, supporting businesses of all sizes in their digital transformation journey. In addition to using embedded AI and allowing for custom AI applications that can access data from S/4 and other SAP products, this presents a major opportunity for businesses to adopt and scale AI. The importance of strategic planning for transitioning to SAP S/4HANA is also underscored, with tools like SAP Readiness Check and RISE with SAP being essential to streamline processes and maintain a clean core.?

Our colleague Mark Chalfen also recently shared his personal observations from Sapphire 2024 here.

In today's rapidly evolving business landscape, the integration of Artificial Intelligence (AI) into organisational processes is no longer a futuristic concept but a present necessity. In our previous articles (AI reflections, AI Role for the Future of ERP's , Embedded GenAI in S/4HANA with Joule, and AI in Financial Decision-Making) we talked about WHAT the opportunity looks like with AI, and in this article we touch on HOW to capitalise on that opportunity. This article explores the essential steps to enable AI transformation, focusing on the integration of AI into existing business processes, the foundational requirements, and the profound impacts on organisational operating models.

Getting familiar with SAP Business AI

In the digital age, integrating AI into business processes is crucial for maintaining a competitive edge. SAP Business AI offers a comprehensive solution that unifies data access across various business domains, enabling organisations to innovate and optimise their operations.

Unified Data Access Across Business Domains:

• Finance: Real-time insights and predictive analytics enhance financial planning and reporting.
• Operations: Intelligent process automation and data-driven decision-making streamline operations.
• Supply Chain: Advanced forecasting, inventory management, and demand planning improve efficiency.
• Manufacturing: AI-driven insights optimise production processes, maintenance schedules, and quality control.
• HR: Personalized AI recommendations transform talent management, employee engagement, and workforce planning.

Innovation Through SAP Business Process (BP) Framework The BTP framework provides a structured approach to managing and optimising business processes. It ensures seamless integration of AI capabilities into existing workflows and enhances user experience with a Fiori front-end, offering an intuitive and responsive design.

Unified Data Agreement via RISE RISE with SAP facilitates business transformation with a single data agreement that ensures data governance, security, and compliance across all integrated systems. This simplifies the integration of AI technologies and leverages data efficiently.

Integration of Leading Large Language Models (LLMs) SAP Business AI integrates leading LLMs such as Anthropic Claude, OpenAI’s ChatGPT, and IBM Watson. These models provide advanced natural language processing capabilities, enabling businesses to automate customer service, enhance decision support systems, and develop intelligent applications.

A Platform for Rapid Innovation Combining SAP Business AI, the BP framework, and access to leading LLMs creates a robust platform for rapid innovation. This platform approach ensures AI is integrated across the organisation, driving cohesive and effective use of technology.

A Case Study: Procurement Transformation with SAP Business AI:

• Simplified Buying Process: On-screen recommendations streamline procurement.
• Data Extraction and Error Minimization: AI-driven automation reduces manual effort and errors.
• Prescriptive Guidance: Data-driven recommendations enhance decision-making.
• Intuitive Sourcing: Facilitates creation of effective sourcing projects based on past successes.

The Pressing Need for AI Integration into Core Processes

As businesses strive for increased efficiency and innovation, embedding AI into core processes is crucial. In 2024, organisations are increasingly recognizing the importance of embedding generative AI (GenAI) into their core processes.?

According to a recent McKinsey Global Survey, 65% of organisations are regularly using GenAI, nearly double the percentage from the previous year. These organisations are deriving business value from GenAI, with reported cost decreases and revenue jumps in business units deploying the technology. GenAI adoption is most common in functions where it can create the most value, such as marketing and sales and product and service development.

The interest in GenAI is global, with more than two-thirds of respondents in nearly every region reporting AI usage. Companies are now using AI in multiple business functions, indicating broader adoption across the organisation. Respondents expect GenAI to lead to significant or disruptive change in their industries in the coming years.?

Over the last 12 months, evidence is increasing that the best adopters of GenAI are getting efficiency, innovation, and competitive advantage in their core processes. As GenAI continues to evolve, embedding it into core processes becomes crucial for staying competitive and achieving sustainable growth.?

The value case story is clear. But how to get there? Over the last 12 months, some key learnings have emerged.

What do the best adopters do well?

1. Establish a Robust Data Platform

A structured data platform is the cornerstone of successful AI integration. To begin with,? GenAI models rely heavily on high-quality data for training and inference. Good data platforms enable data preprocessing, including tasks like cleaning, feature engineering, and normalisation - making GenAI process execution simpler. A robust data platform ensures scalability, allowing organisations to handle large datasets efficiently. Efficient data processing accelerates model development and deployment, contributing to RoI.?

We believe a good data platform must be capable of aggregating data from various sources, ensuring data quality, and providing a unified view of the organisation’s information to maximise GenAI value. Once this is achieved, it will support the seamless deployment of generative AI models that can analyse and interpret data to drive decision-making.

2. Robust and well-integrated processes

Consistent, repeatable, system based processes are a key ingredient to getting value from AI investments. Deploying AI over sub-optimal processes can seem appealing but offers incremental value as a point solution only, not the step change that AI should be. Good processes allow GenAI to be deployed faster, used more effectively, used in more complex and higher value scenarios and rolled out at scale faster.

Robust processes also drive data quality and governance which as covered above is a cornerstone for success. Additionally they enable the right controls, which is crucial not just for process execution but also for responsible AI adoption. Slogans such as “fit to standard” or “clean core” have not always landed with clients but GenAI value realisation could make this a reality.

3. Build your AI-first business operating model

The integration of AI into business processes will necessitate a significant shift in organisational operating models. As transactional activities become automated, the demand for roles focused on soft skills such as persuasion, influence, and communication will increase. Organisations will need to repurpose their workforce towards these higher-value activities, ensuring employees are equipped with the necessary skills to thrive in an AI-enhanced environment.?

This transformation will also affect the structure and size of teams, potentially reducing the number of employees required for transactional tasks while increasing the need for roles focused on strategy and innovation. Companies will need to invest in continuous learning and development programs to upskill their workforce and prepare them for the evolving demands of the workplace.

How you should think about your AI transformation

1. Define your strategy

With 700+ AI providers on the market and leading tech organisations increasingly championing their own AI solutions, our clients have key choices to make:

• Can we implement point solutions or do we need AI embedded throughout core processes?
• Should we buy, build or consume as a service?
• What is the right priority order for us to maximise RoI?
• Which vendor(s) should we partner with?

2. Maximise Vendor-Embedded Solutions for core processes

Many AI solutions are now embedded within vendor platforms, offering specialised capabilities tailored to specific business functions. A solution specifically designed by the vendor for their standard process offers significant advantages to clients. For example, with SAP Joule the infusion of AI will be revolutionary. By the end of the year, 80% of the most common tasks performed by SAP's 300 million end users will be managed by Joule, potentially increasing productivity by 20%. Maximising use of Joule therefore makes complete sense.

We are reaching a maturing AI market where the major tech players are either acquiring boutique AI solutions or partnering with other major players. Forrester refers to this as “Goodbye, best-of-breed and welcome suites (again)”.This is very similar to the market behaviour with previous tech innovations - think of smartphones, cloud, big data, analytics for example.?

At Sapphire, SAP announced a series of expanded partnerships to help companies make the most out of this decisive AI moment. The expanded partnerships include deals with Amazon Web Services (AWS) and Google Cloud. Klein also announced the integration of Microsoft 365 Copilot with Joule, delivering what he described as a “truly differentiated employee experience.”?

2. Supplement them with productivity assistants

Even the best run processes need to account for human behaviour. Regardless of how good an ERP platform’s self-service analytics can be, users will still resort to moving data into Excel, Word and PowerPoint for “last mile” or ad-hoc processes. That’s why SAP is integrating Joule with Microsoft Teams and CoPilot to keep processes running “in the system” even when users are out of the system. The integration combines enterprise data from SAP with contextual knowledge from Microsoft 365, including tools like Microsoft Teams for notifications, Outlook for calendar and email, and Word for documentation. It is said that users will benefit from a unified experience, regardless of which copilot they use.

Close integration between the major vendors in your IT ecosystem creates opportunity for you to get a harmonised user experience, better integration across applications and more value from your (expensive) subscription fee. Making the various AI copilots talk to each other could be the interface design approach of the future.

3. Buy best-of-breed for niche or high complexity areas

In many cases where there are specialised needs, purchasing proven tools can be more advantageous than building them, enabling rapid scaling and access to advanced AI capabilities. For example, using specialised tools such as Harvey for Tax or Truillion IFRS 16 for deep content contract management and extraction can be more efficient.?

While generative AI is highly useful, it is just one of many tools available. There are still scenarios where deploying machine learning or other structured analytics tools may be more effective than generative AI.

4. Build your own for truly differentiating areas

The knowledge, skills and investment required to develop proprietary AI solutions is significant, and there are few examples of organisations who have truly cracked this yet. There is also a concern that a proprietary AI solution will become obsolete very quickly as (for example) a vendor AI solution matures. Therefore it is sensible to reserve this option only for truly differentiating use cases.

So in conclusion:

1. AI without trusted data and good processes will be incremental at best
2. Multi-AI will soon be complex to manage so the e2e process model is the unifying factor
3. Make the AI tools talk to each other to enable smooth processes across applications
4. When building, the focus is the LLM but this is by far the easiest aspect compared to getting people on board

What this could look like if you get it right

Finance

Sapphire had a number of presentations on the future of finance with Joule. The fusion of generative AI and automation was a focal point, promising substantial benefits for businesses. Joule AI agents are poised to be central in achieving automation driven by generative AI by automating and streamlining an increasing amount of GBS services.?

These AI business agents can be applied in numerous areas, including sales support, customer care, procurement administration and business reporting. Realising these applications requires seamless semantic and technical integration between business systems, processes, and AI agents, with an emphasis on security and accuracy. To address this, the SAP Innovation Center Network Walldorf has developed a prototype called the Business Agent Foundation on the SAP Business Technology Platform.

The Business Agent Foundation prototype aims to facilitate the efficient implementation of AI-driven business scenarios by offering reusable AI business agents as a service. It provides various integration and development tools, allowing AI agents to be easily incorporated into existing and new business applications. This innovation equips agents with the necessary skills to handle enterprise processes and tasks while comprehending business context and semantics.

This is fascinating in itself. But image this working alongside the other GenAI applications to deliver an e2e finance process. One example of this coming to life is Accounts Receivable:

What could Joule do:

Automation of all AR tracking, administration, follow up and reporting using business agenta

Where can best of breed be used:

Specialised tools such as HighRadius for complex collections and disputes case management, along with expert tools such as Truillion for complex revenue recognition postings or accounting adjustments

How can Microsoft CoPilot help:

Use of MS Teams notifications to coordinate globally distributed teams, including creating follow up and reminder notifications in Outlook

And how does this come together:

Deep analysis and on-demand management recommendations in a MS Azure Data Lake combining ERP, non-ERP data and AI assistant data

In Conclusion

Enabling AI transformation requires a comprehensive approach, encompassing the establishment of a robust data platform, leveraging generative AI capabilities, and utilising vendor-embedded solutions. By embedding AI into core business processes, organisations can automate routine tasks, reduce cycle times, and derive actionable insights. This transformation will have far-reaching impacts on operating models, necessitating a shift in workforce focus towards higher-value activities and strategic initiatives. As AI continues to evolve, its integration into business processes will undoubtedly become a critical driver of organisational success and competitiveness.