If you haven’t heard about AI agents, you might want to check if your Wi-Fi’s working, or maybe you really have been living under a rock. In just a short time, these digital co-workers—or assistants, copilots, and other nicknames—have taken center stage in tech. And the hype is real. Expectations for what AI agents can do are sky-high; some imagine they’ll soon run the whole show, making decisions for us while we sip our coffee. But do we really want them to do everything on their own? And can they actually do that?
As companies race to implement this new technology, they’re discovering it’s not all as smooth as envisioned. Following a recent Gartner report, high costs and fuzzy business value are creating speed bumps. As it turns out, the technology itself isn’t the problem, it’s how and why we use it. Like any shiny new gadget, AI agents only matter when they solve real problems that make people’s lives easier. So, what kinds of issues are they good at tackling? And how do we make sure we’re designing systems that serve actual humans and are not just chasing the latest tech trend?
That’s where things get interesting: deciding when you truly need an agent, how much freedom it should have, and what challenges and tasks it’s meant to address all while ensuring it genuinely helps people, instead of just ticking the “we use AI” box. The real magic happens when humans and agents team up, working side by side for the best results. How do we make the most of this human-agent partnership?
If you’re looking for a practical way to get started, the SAP AppHaus Joule Agent Discovery and Design workshops offer a hands-on approach to help tackle these exact questions. With a blend of human-centered design methods, these workshop formats put people first, working to ensure agentic systems aren’t just flashy but genuinely useful.
Want to try it out for yourself? Here’s how you can run your own workshops and define impactful agentic systems.
A toolkit to build human-centered agentic solutions
The Joule Agent workshops are offered as two different formats, each designed to guide you through a different stage of building effective agentic systems: the Joule Agent Discovery workshop and the Joule Agent Design workshop. Together, these workshops provide a hands-on, human-centered path for creating AI agents that can truly deliver value.
First stop: Joule Agent Discovery workshop
The Joule Agent Discovery workshop is a structured approach to uncover the most valuable opportunities for agentic technology. It focuses on real-world challenges and identifying where automation can make the biggest impact. In two to three hours, participants dive into questions such as: What specific inefficiency or challenge needs solving? What could be automated? Who would benefit most from automation? What needs to be achieved with the automation? How complex and variable is the problem at hand?
The workshop also introduces participants to agentic technology and examines how much the selected challenges would benefit from it. By the end of this workshop, participants identify one or more high-value use cases that are well-suited to agentic technology. This helps ensure that efforts are focused on meaningful improvements rather than adopting technology for its own sake.
Second stop: Joule Agent Design workshop
Next is the Joule Agent Design workshop, which brings together those closest to the process—end users and business experts—to define the details of the agent: its responsibilities, required skills, and how it will collaborate with people. The workshop follows a practical structure:
- Define the focus area: Clarify what target users need to achieve within the selected process and identify which aspects would benefit most from automation.
- Identify tasks to delegate: Use the metaphor of “hiring a super-specialist” to decide which responsibilities should remain with people and which can be assigned to agents. Exercises help determine how many agents are needed, the risks of automating certain tasks, and where consistency versus autonomy is required.
- Describe the super-specialist job: Draft a job description for each agent, outlining necessary skills and responsibilities.
- Instruct the super-specialist: Define the instructions or workflow, including information requirements, decision points, and where human involvement is needed.
By the end of the workshop, each agent is described in detail, including its tasks, required knowledge and tools, and an initial set of instructions. This forms the foundation for configuring the agent’s system prompt.
The workshop material also offers guidance on structuring the system prompt based on the gathered information, ensuring a smooth transition from workshop insights to practical implementation. The entire process is designed to be completed in a single day and can be conducted virtually in Mural.
Learning how to run these workshops
To help ensure that anyone can confidently run these workshops—no advanced degrees or secret codes required—a set of self-paced courses are available and can be completed at the individual pace of the learner:
- Discovering High-Value Opportunities for Agentic AI: This course offers a comprehensive introduction to the Joule Agent Discovery workshop. It provides a clear, step-by-step guide, explains the exercises in detail, and gives practical advice on how to facilitate effective sessions. You’ll gain the skills to identify agentic opportunities and successfully lead your team through the process.
- Spotting Agentic Opportunities in Practice: This short webinar also centers on the Joule Agent Discovery workshop, but it specifically highlights a practical method for assessing the agentic potential of automation ideas. Consider it your quick reference for making informed decisions about automation.
- Designing Agentic Systems with a Human-Centered Approach: The latest addition to our curriculum, this course walks you through the Joule Agent Design workshop step by step. It covers each exercise, shares real-world examples, and offers facilitation tips. You’ll also learn how to adapt the workshop format for various time constraints and organizational needs.
All the resources required to facilitate these workshops are freely available on the innovation toolkit for AI website. Learners can simply visit the site, explore the materials, and start their agentic journey with confidence to turn ideas into new useful, human-centered AI solutions.
Karen Detken is an expert user experience designer at SAP AppHaus.
