Building AI agents
About this training
Learn to design, build and test a practical AI agent with ChatGPT. You can also apply the framework in Claude, Copilot and other AI platforms.
A chatbot answers your questions. An AI agent helps you carry out tasks.
Building an AI agent means creating a smart assistant that processes information, follows defined steps, prepares decisions and supports recurring tasks.
This practical AI course teaches you how to build a simple AI agent yourself. You start with a specific task or workflow: what do you want to simplify, what information does the agent need, and where does human oversight remain essential?
Together, we build a working AI agent using a practical example: an email agent that helps you process, forward and reply to emails. Along the way, you learn a clear framework that you can apply to other tasks in communication, marketing, sales, HR, administration, finance or internal services.
We use ChatGPT for the practical exercises. However, the principles and framework also apply to other AI environments, including Claude and Microsoft Copilot.
By the end of the course, you will have built a working AI agent that you can adapt to your own professional context.
What is an AI agent?
A traditional chatbot responds to a question. An AI agent goes a step further. You give it a clear role, an objective, relevant information and a process to follow. This creates a smart AI assistant that supports you consistently with a specific task.
For example, an AI agent can:
- analyse and structure incoming information;
- summarise and process documents;
- prepare emails or other responses;
- categorise and route requests;
- gather information for a report;
- prepare content for different audiences;
- turn meeting notes into action points;
- support recurring administrative tasks;
- prepare proposals or decisions;
- carry out checks against defined criteria.
An AI agent does not simply take over all the work. You decide what it is allowed to do, which information it uses, and when a team member must check or approve its output.
Building AI agents with ChatGPT, Claude or Copilot
During the practical exercises, we use ChatGPT as our main working environment. However, the principles of designing an effective AI agent apply more broadly. In Claude and Microsoft Copilot, you also need to define your assistant’s role, context, knowledge sources, steps, boundaries and review points clearly.
Would you prefer to develop your skills in one specific AI tool first? Explore our:
- ChatGPT course for professionals
- Claude AI training for professionals
- Microsoft Copilot training for the workplace
In-company training: building AI agents
This course is also available as in-company AI training. The practical case and exercises are then tailored to your organisation’s processes, AI tools and working environment.
Before the course, we discuss:
- which AI environment your organisation uses;
- which tasks or processes participants want to improve;
- which security and privacy rules apply;
- which documents or practical cases can be used during the course;
- which outcomes your organisation expects after the training.
This allows participants to work on familiar situations and achieve usable results more quickly.
Do you use Claude, Copilot or another AI tool?
We build the agent in ChatGPT, but the approach is not tied to a single tool. Choosing a task, setting boundaries, writing instructions, testing and building in oversight: the same approach applies in Claude, Copilot and other AI platforms. The buttons change; the logic stays the same.
Would you also like to explore a specific tool in more depth? Take a look at our Claude AI training or our Copilot training.
Frequently asked questions about building AI agents
What is the difference between a chatbot and an AI agent?
A chatbot answers your questions. An agent carries out a series of steps, retrieves information and processes it within the boundaries you define in advance. A person remains in control of important decisions.
Which subscription do I need for this course?
You need a paid subscription to the tool you use: at least ChatGPT Plus, Claude Pro or Copilot Premium. Arrange this before the course starts so you can get to work immediately.
Do I need programming skills to build an AI agent?
No. You work with everyday language and clear instructions. However, some previous experience with an AI tool is helpful.
Do I have to use ChatGPT, or does this also work in Claude or Copilot?
We build the agent in ChatGPT. The framework also works in Claude, Copilot and other AI platforms.
Can I build an agent for a task other than email?
Yes. The email agent is the example used during the course. You can then apply the same steps to tasks in areas such as accounting, marketing or customer service.
Is it safe to let an AI agent work with confidential data?
That depends on how you configure the agent and which data you provide. That is why we dedicate a complete module to privacy, error handling and approval by a team member.
Can my team take this course as in-company training?
Yes. We tailor the course to your organisation’s tools and tasks. Contact us for a tailored proposal.
Other AI courses
Explore our full range on our AI training courses page.
Ready to move beyond using AI and start putting it to work? Register for the course or request an in-company training proposal.
Program
This practical course takes you step by step through the complete process of designing, building, testing and improving an AI agent.
- Discover what an AI agent actually does. Learn the difference between a chatbot, an AI assistant, a workflow and an AI agent. Explore which tasks an agent can support, which decisions it can prepare, and where human oversight remains essential.
- Choose a suitable task or workflow. Learn to identify recurring tasks, established rules, information sources and decision points. Then map the process inputs, individual steps, desired output and review points.
- Design your own smart assistant. Define the agent’s role, who it works for, the result it should deliver, and what it is and is not allowed to do. Specify when it should stop, request additional information or hand the task over to a team member.
- Build a working AI agent in ChatGPT. Turn your design into clear instructions and add relevant context, documents and knowledge sources. Using a shared practical case, build an agent step by step to support a recognisable professional task.
- Learn how to make the agent work with information and applications. Discover how an agent can retrieve, structure, process and pass on information. We explore how agents can work with documents and applications such as email, calendars, spreadsheets and forms, depending on the platform and available access permissions.
- Test and improve the agent. Test your agent with realistic situations, incomplete information and exceptions. Analyse what goes wrong and refine the instructions, information sources and review points until the output is sufficiently reliable and useful.
- Build in safety and human oversight. Learn to account for privacy, confidential information, hallucinations and incorrect output. Decide when a team member needs to check, approve or intervene, and how to monitor the agent’s performance.
- Apply the framework to your own work. Use the same method for a task within your role or organisation. This could be an agent for communication, marketing, sales, HR, administration, finance, reporting or internal services.
You leave with a working agent that you have built and tested yourself, along with an approach you can apply to future tasks.
Target group
This course is designed for professionals who already use ChatGPT or another AI tool regularly and want to take the next step: moving from individual prompts to a reusable AI agent for a specific task or workflow.
The course is suitable for experienced AI users who want to:
- get more out of ChatGPT than one-off questions and requests;
- organise recurring tasks more effectively;
- learn to design, build and test their own AI agent;
- integrate AI into their workflows in a controlled way.
Basic experience with a generative AI tool is required. If you have barely used ChatGPT or a similar tool, we recommend starting with our ChatGPT course for professionals.
Bring your laptop so you can take part in the practical exercises immediately. You also need a paid subscription to the AI tool you want to work with: at least ChatGPT Plus, Claude Pro or Copilot Premium.
Approach
This interactive workshop is highly practical. You will immediately start working on concrete assignments and exercises.
The training combines theory, demonstrations, hands-on practice and extensive interaction. You will learn how to apply generative AI tools to your own objectives, build AI-driven chatbots live, and experiment with image recognition and text generation.
Be sure to bring your laptop to fully benefit from this session.
So be sure to bring your laptop so you can get started right away in this workshop.
Trainers
Results
By the end of this course, you will have:
- built a working AI agent or a concrete prototype;
- acquired a reusable framework for designing other agents;
- identified tasks and processes suitable for AI support;
- learned how to test and improve agents systematically;
- developed an understanding of privacy, reliability and human oversight;
- created an initial design or action plan for an application within your own working environment.
You leave with more than inspiration: you gain a concrete method that enables you to continue building independently.
Register
Choose the desired date and location for this training.
Gent - Expert Academy
From
€ 495,00(VAT excluded)
Max. 8 participants
Antwerpen - Expert Academy
From
€ 495,00(VAT excluded)
Max. 8 participants
