Introduction
As generative AI continues to transform business operations, AI agents are emerging as key players, enhancing collaboration and redefining efficiency. Building on my previous blog post, I now explore three compelling use cases of AI agents in action to demonstrate how AI agents streamline customer service, cut the time it takes to complete customer dispute resolutions time and how to easily build and customize AI agents with no-code workflows.
What are AI Agents? Watch the video from Jonathan to find out more.
Before I deep dive into AI agent use cases, let’s first clarify when it actually makes sense to build an AI agent use case?
Implementing AI agents in business processes can significantly enhance efficiency and decision-making when applied to suitable tasks. Key criteria for determining the appropriateness of AI agents include:
- Processing Unstructured Data: AI agents excel at analyzing and organizing unstructured data – such as emails, documents, or tickets – transforming it into structured formats. This capability enables businesses to extract valuable insights and automate subsequent actions based on the classifications. For example, AI can automatically categorize customer inquiries and route them to the appropriate department, creating case documentation and subsequent actions and thus enhancing response times and customer satisfaction.
- Enhancing Decision-Making with Data Analysis: AI agents can process large volumes of data to identify patterns and trends that might be overlooked by human analysis. This supports informed decision-making in areas like market analysis, risk assessment, and strategic planning. By providing data-driven recommendations, AI agents empower businesses to make proactive and effective decisions and make business priority the key criteria for the users’ attention.
- Automating Complex Repetitive Tasks with Contextual Understanding: Unlike traditional Robotic Process Automation (RPA) systems that handle straightforward, rule-based tasks, AI agents can manage repetitive tasks requiring nuanced understanding and minor contextual decisions. This allows for the automation of long-tail business processes previously unsuitable for automation. For instance, AI agents can interpret customer feedback, discern sentiment, and initiate appropriate follow-up actions, thereby enhancing customer engagement and operational efficiency.
These criteria should of course not be considered exhaustive, but by evaluating your business processes against these criteria, organizations can strategically implement AI agents to automate tasks, enhance accuracy, and drive growth.
Let’s have a look at some agentic use cases and how they are transforming organizational efficiency across lines of businesses
How AI Agents Streamline Customer Service
Normally, when a customer service ticket comes in, customer services teams start manually classifying the case using basic routing rules – often spending valuable time just getting tickets to the right team. Next, service representatives spend significant time searching documentation and reviewing similar cases – often across multiple systems. When they reach a resolution, they draft responses and might update the knowledge base if they remember and have time. And here's where missed opportunities frequently occur – in the sales handoff phase. When potential sales opportunities are identified and created, unfortunately, sales teams often have to ask customers the same questions again because context gets lost in the handoff. This process works, but many customers find it time-consuming, nerve-racking, and prone to mistakes.
Now, let’s explore how this process is being transformed with AI agents. Each stage is now enhanced with intelligent automation while the customer service team remains in control. The Classification Agent handles smart categorization and instant routing based on business rules. It's like having an expert agent who knows all your routing protocols working 24/7. To find solutions, the Q&A Agent automatically searches the knowledge base and matches solutions – imagine having the most experienced agent's knowledge instantly available. In the resolution stage, the Knowledge Agent not only documents the solution but intelligently captures customer needs and usage patterns. This is crucial because it feeds into the next stage. Finally, the same Q&A Agent that helped with service, now provides immediate answers to any product or upgrade questions included in the opportunity notes. No more repeated questions or lost opportunities.
What's powerful about this new process is that it maintains all the context and customer intelligence throughout the journey, while automating the time-consuming tasks that previously slowed teams down. Teams can focus on building customer relationships while the AI handles the heavy lifting of information gathering and routing.
These AI agents are delivered through the CX AI Toolkit and work out of the box with SAP Sales Cloud and SAP Service Cloud, enhancing the processes and tools customers already use. CX agents are generally available today, and they offer new workflows to enhance the typical sales and service process within your SAP environment.
Check out the demo to see what this looks like within the SAP Service Cloud: https://www.sap.com/assetdetail/2025/01/b8dba82e-f27e-0010-bca6-c68f7e60039b.html
How AI Agents Cutting Dispute Resolution Time
Let’s explore another scenario which involves multiple agents simplifying the most complex tasks and processes using AI. You will see how proactive agents can monitor and react faster than any human.
Picture this: you’re an accounts receivable clerk, responsible for receiving and processing customer queries daily. Balancing the need to collect money with maintaining customer satisfaction is a tedious, time-consuming process. Now, the good news is: agents can help collect this money more quickly.
In this scenario, let's explore how a customer experience agent communicates directly with the customer, while a cash flow agent coordinates the entire process. These agents operate seamlessly within the SAP system and collaborate in the background with Microsoft Copilot.
An accounts receivable clerk receives an urgent customer issue via a Microsoft Teams notification. The integrated AI system immediately jumps into action, outlining the steps the agents have taken: reading the customer’s email, retrieving pertinent information, creating a dispute proposal, and more. The clerk can easily view the invoice by accessing it in SAP S4/HANA Cloud.
To understand the concrete next steps, the clerk asks the SAP AI copilot, Joule, to provide a detailed update. The collection agent communicates with multiple other agents – customer service, financial accounting, and shipping and receiving – to gather the necessary information, and then the final recommendation is displayed. This process would usually take hours, and now requires only a few seconds.
Furthermore, Joule not only presents the resolution but also the logic, reasoning, and sources behind it. The clerk can see the various data sources, objects, and documents that were used as part of this orchestration to come up with the resolution. Additionally, Joule provides two alternative suggestions, offering a comprehensive view of potential solutions. Let’s suppose the clerk decides to accept the resolution and to select goodwill option B. With a simple command, the system puts the decision into action, with the agents working together to create and send the necessary email using a large language model (LLM). This being said: The human always has the final say. Now looking at the dispute case that's been created in the system it’s documented that the disputed amount is settled and the dispute is closed – what a success!
After making sure the customer stays happy, the next task is addressing the issue caused by the third-party shipping company that failed to deliver on time. By asking Joule for guidance, the clerk can quickly determine the best course of action.
Joule's agents analyze the situation and conclude that the shipping company clearly violated the service level agreement. The agents promptly recommend opening a dispute case. With just one click, the dispute case is created and sent off to the shipper.
A process that would usually take days or weeks has been solved in just a few minutes – all thanks to the collaborative efforts of the AI agents. This level of efficiency and effectiveness is what makes working with these agents so powerful and beneficial.
Watch the video for more: https://www.youtube.com/watch?v=1Ku7J4XIfE8
Easily Build and Customize AI Agents with No-Code Workflows.
Every business and industry has unique requirements, and creating custom agents that suit your organization is now straightforward using Joule Studio in SAP Build.
For instance, if the business is growing and you need to renegotiate supplier contracts for increased volume, you can create a new Joule agent to develop strategies that support your negotiations.
Let me walk you through the steps to create a new agent from scratch!
To begin, you'll define the agent’s purpose, for example, helping procurement managers negotiate better pricing and terms. Joule guides you through this process, and you’ll name your agent something fitting, like "Pricing Negotiator." Joule then takes us to the build page and fills out the basic information. It analyses the description and from this generates an avatar and a banner image.
Next, the bounds of the agent’s role can be set by assigning it to a business process, using the to identify the correct sub-process within the source-to-pay process. Collaboration is crucial, so the Pricing Negotiator is connected to other relevant agents. Adding third-party agents like Microsoft 365 Copilot is also straightforward, requiring just one simple click.
Data is essential to ensure the Joule agent is grounded in business context. Using the SAP Business Data Cloud and SAP Knowledge Graph, relevant data products such as contracts, invoices, and payment history can be easily identified and added. Externally sourced information, like third-party benchmarking, can also be seamlessly integrated.
The Pricing Negotiator is then equipped with the necessary tools by giving it access to functional libraries of Joule skills, such as invoice management and supplier search.
We need to define when the Pricing Negotiator will be active – this is done via triggers. We have multiple options here, the default is calling it directly from Joule but we could create a schedule or link it to another process.
Triggers are defined to determine when the agent will be active, and communication channels for human collaboration, such as Microsoft Teams, are set up. Joule summarizes the agent’s objectives and tools, allowing for the addition of constraints, such as not negotiating with suppliers overdue for more than 90 days until deliveries are resolved.
To enhance performance, a reasoning model is used to help the agent make more nuanced judgments. Finally, the agent is tested by entering a prompt like “Look up the last RFQ issued to our supplier Best Run and create a negotiation strategy.” The Joule agent starts to reason, plan, and act – using all the business context, tools, and other agents as needed to decide on the best strategy, including key negotiation tactics, risks, and even alternatives!
And right at the end of it, there’s a concise summary.
Creating custom Joule agents is a straightforward process that significantly enhances business operations by making teams more efficient and streamlining workflows.
Watch the Joule Agent Builder demo here: https://www.sap.com/assetdetail/2025/02/4212ffbc-f27e-0010-bca6-c68f7e60039b.html
Let’s now briefly touch on Responsible AI, the core of SAP’s mission.
As you read earlier, the human always has the final say. Responsible AI is a critical priority for SAP because it aligns with the company’s broader values of trust, transparency, and sustainability, which are central to our role as a leader in enterprise software. Consequently, Joule Agents are designed with a human-in-the-loop (HITL) framework, ensuring that human oversight is always maintained in AI-driven operations. This approach allows humans to review and validate AI decisions, achieving transparency and enhancing reliability. Users have the final authority to approve or modify AI recommendations, reducing errors and incorporating expert judgment.
AI Ethics and Governance for Agentic AI emphasize several crucial areas:
- Human Oversight: Ensuring human judgment is integral to decision-making.
- Transparency and Explainability: Making AI processes clear and understandable.
- Fairness and Non-Discrimination: Preventing biased outcomes, especially in applications affecting individuals.
The HITL approach is essential in use cases that:
- Process Personal Data or Personal Identifiable Information
- Potentially Impact Individuals Negatively
- Operate in High-Risk Domains (e.g., HR, critical infrastructure, law enforcement).
By adhering to these principles and structures, businesses can ensure that AI agents are used responsibly and ethically. The integration of responsible AI practices includes human oversight, collaboration, and adherence to ethical guidelines, ensuring efficient, ethical, and transparent AI operations.
By now, it’s clear that AI agents are intended to amplify and accelerate the work of humans. As they reshape the way we work, they accelerate the execution of complex tasks and augment decision-making, allowing human workers to focus on more strategic activities that require human insight and creativity. As with any technological shift, the key is to embrace AI as a collaborator, leveraging it to enhance productivity and innovation, rather than viewing it as a threat. Businesses that invest in upskilling their workforce will unlock the full potential of AI while ensuring employees remain at the center of decision-making. This relationship between humans and AI agents fosters productivity and innovation, reflecting a forward-thinking approach to modern work environments.
Speaking of upskilling! Make sure to engage with our newly released Learning Journey “Boosting AI-Driven Business Transformation with Joule Agents”. After completion, you will be able to understand and utilize Joule Agents to drive business transformation, enhance decision-making, and integrate AI capabilities.
