Webinar teaches users to build custom assistants with Claude
A new training session from c't explains how to move beyond simple chat prompts to create personalized AI workflows with specific files and rules.

The short version
- Users can create separate 'Projects' to organize different work areas and keep the interface clean.
- The training covers how to teach Claude a specific writing style and provide it with private knowledge files.
- The 'Cowork' function allows the AI to complete multi-step tasks autonomously without constant prompting.
- The session includes a dedicated segment on minimizing security risks when the AI accesses web data or files.
- No programming knowledge is required to attend the session.
Why it matters
Most people use AI as a simple search engine, but this training shows how to turn it into a specialized team member. By providing context and rules, users can automate complex, multi-step office tasks more reliably.
Moving beyond simple chat
Many people interact with AI assistants by asking a single question and then closing the window once they receive an answer. While this works for basic queries, it does not take full advantage of what tools like Claude can actually do for professional work. As heise.de reported, the goal of the upcoming webinar is to show how users can provide much more context to the chatbot.
By giving the assistant specific files, folders, and clear rules, the AI becomes more useful for complex tasks. Instead of just answering questions, it can work within specific documents like Word or Excel files. It can also analyze shared folders to perform tasks that require multiple steps of reasoning or data processing.
The webinar, titled 'Claude in der Praxis,' aims to bridge the gap between casual use and professional integration. It is designed for both beginners and those who have already used other models like ChatGPT or Gemini. The training focuses on how Claude handles specific work contexts differently than its competitors.
Setting up projects and styles
A core part of the training involves learning how to organize work using the 'Projects' feature. This allows users to create separate workspaces for different tasks, which helps keep the sidebar organized and manageable. By isolating different work areas, the user can ensure the AI stays focused on the relevant information for that specific project.
Users can also teach the assistant a specific 'Style' to match their own writing tone. This ensures that the output produced by the AI remains consistent with the user's personal or corporate brand. Additionally, the webinar covers how to upload 'knowledge files' so the AI has a dedicated source of truth to reference during interactions.
Jo Bager, a c't editor, will demonstrate which tool is best for which situation. He will explain when it is better to use a project, a style, or a saved memory. This distinction helps users choose the right configuration to get the most accurate results while maintaining a clean workspace.
Automating multi-step tasks
One of the more advanced topics covered is the 'Cowork' function, which enables the AI to handle multi-step instructions. In a standard chat, a user might have to prompt the AI for every single step of a process. With Cowork, the assistant can work through a complex order more autonomously without needing constant human intervention.
The webinar will show how users can monitor these individual steps as they happen. If the AI makes a mistake or produces an unwanted result, the user can correct it at that specific point in the chain. This level of control allows for more complex workflows while still giving the human user the final say on the output quality.
Furthermore, the training explains how to save these recurring workflows as 'Skills.' Once a multi-step process is perfected, it can be reused easily. This turns a one-off task into a repeatable capability, making it much more efficient for daily office routines that involve repetitive data handling or drafting.
Security and limitations
As AI assistants gain the ability to read files and browse the web, they naturally create new security risks. The webinar includes a dedicated section on how to manage these risks effectively. Bager will show specific settings that users can adjust to keep their data and connections as secure as possible.
It is also important for users to understand where the automation reaches its limits. The training will highlight signs in the final output that indicate the AI might be struggling or hallucinating. Knowing these boundaries helps users decide when they need to step in and take over the task manually to ensure accuracy.
The session is designed to be accessible to everyone, meaning no programming knowledge is required to participate. Participants are encouraged to have a free Claude account and a current web browser. This allows them to follow along and try the steps immediately after the webinar concludes on December 8, 2026.
Practical application for work
The ultimate goal of the training is to move the user from a simple login to a fully integrated work routine. It starts with the basics, such as creating an account, choosing a tariff, and adjusting privacy settings. These foundational steps ensure that the user is set up correctly before they begin building more complex assistants.
By the end of the session, participants should understand how to move beyond the 'question and answer' cycle. They will see how to build a system where the AI understands their specific documents, follows their unique writing style, and executes multi-step commands. This transition turns the AI into a functional assistant rather than just a novelty tool.
The webinar provides a structured path for those who want to be more productive with AI. By focusing on concrete example tasks, it gives users a blueprint for their own daily work. It emphasizes that the real power of these tools lies in the context and rules the human user provides to the system.
Next steps for users
The webinar is scheduled to run from 15:00 to 17:30 on December 8, 2026. It costs 69 Euro and is aimed at people who want to see practical applications of Claude in a professional setting. The session will provide a clear roadmap for anyone looking to optimize their workflow with AI technology.
For those interested, the seminar website provides all the necessary details for registration. Because the training is hands-on, having a Claude account ready is highly recommended. This allows participants to replicate the demonstrated steps, such as creating projects or uploading files, as soon as the webinar ends.
The training highlights that while AI is powerful, it requires proper setup to be truly effective. By learning how to manage styles, knowledge files, and multi-step tasks, users can significantly reduce the time spent on manual drafting and data organization. It is a practical guide for turning a chatbot into a customized work companion.
Summary of key features
The webinar covers several specific features that distinguish Claude's professional capabilities. One of the primary features is the ability to work with Word and Excel files directly. This allows the AI to interact with standard office documents rather than just processing plain text inputs from a chat box.
Another key feature is the ability to analyze entire folders of information. This is particularly useful for researchers or project managers who need the AI to have a broad understanding of a specific topic. By referencing these shared folders, the AI can provide answers that are grounded in the user's specific data rather than general internet knowledge.
Finally, the training emphasizes the importance of 'Prompts' and 'Styles.' A prompt is the specific instruction given to the AI, while a style is a saved set of rules for how the AI should behave. Mastering these two elements is what allows a user to move from generic AI responses to highly tailored, professional outputs that meet specific requirements.
What to expect from the training
The webinar is structured to take the user from the very first login to a fully functional assistant. It begins with the basics of account creation and model selection. This ensures that even those who have never used Claude before can follow along without feeling overwhelmed by the initial technical setup.
As the session progresses, it moves into the more nuanced aspects of AI management. This includes setting up projects to keep different work streams organized. It also covers the more advanced 'Cowork' function, which is designed for tasks that require the AI to perform several actions in a row without being prompted for every single one.
The training concludes by addressing the most important concerns for many professionals: security and reliability. By showing how to limit risks and identify when the AI is reaching its limits, the webinar provides a balanced view of the technology. It aims to give users the confidence to use these tools in a real-world work environment safely and effectively.
Comparison with other tools
For users who are already familiar with other AI models, the webinar will highlight what makes Claude different. While many people use ChatGPT, Gemini, or Copilot, Claude offers specific ways to handle context and files that might be unique to its architecture. The training will point out these differences to help users decide which tool fits their specific needs.
One of the main points of comparison is how Claude handles 'knowledge files' and 'styles.' While other tools may have similar features, the webinar focuses on how to use these specifically to create a cohesive assistant. It shows how to keep the interface clean while still providing the AI with a large amount of background information.
The goal is not to replace other tools but to show the specific strengths of Claude for professional tasks. By understanding how to set up projects and multi-step workflows, users can see where Claude might be a better fit for certain types of office work. The webinar provides a practical look at these capabilities through real-world examples.
Practical workflow examples
The webinar will use concrete example tasks to demonstrate how to use Claude in a daily work routine. Instead of abstract theories, Jo Bager will show how to move from a first chat to a set of automated tasks. This includes showing how to give the AI a multi-step order and watching it complete the steps independently.
One example might involve taking a large amount of information from a shared folder and turning it into a specific report. Another might involve teaching the AI to write emails in a very specific corporate voice. These examples help users visualize how the technology can actually save time on repetitive tasks that currently take up a lot of their day.
By seeing these tasks performed in real-time, users can learn how to spot the limits of the AI. They will see where the automation might fail and how to intervene to correct the output. This practical approach ensures that the knowledge gained from the webinar can be applied immediately to the user's own specific job requirements.
Managing AI limitations
A significant portion of the webinar is dedicated to understanding what the AI cannot do. Even with advanced features like Cowork, there are points where the automation might struggle. The training will show users how to recognize these limits by looking closely at the final results produced by the assistant.
When the AI reaches its limits, it might provide incorrect information or fail to follow a complex logical step. Bager will show how to identify these moments so that the user knows when to take over. This is a crucial skill for anyone who wants to use AI reliably in a professional capacity where accuracy is important.
Understanding these boundaries helps prevent errors from going unnoticed. By learning to spot the signs of AI struggle, users can maintain better control over the final product. The webinar provides the tools to balance the speed of AI with the necessary oversight of a human worker to ensure high-quality results.
Security and data privacy
Security is a major concern for anyone considering using AI for work tasks. When an assistant is allowed to read files or search the web, it can open up new ways for data to be accessed or leaked. The webinar addresses these concerns directly by showing how to configure the AI to minimize these risks.
Bager will demonstrate specific settings that users can use to protect their information. This includes understanding what data the AI can access and how to limit that access to only what is necessary for the task at hand. These settings are vital for anyone working in an environment with strict data privacy requirements.
The goal is to enable users to use the most powerful features of the tool without compromising safety. By following the security guidelines provided in the webinar, users can build assistants that are both highly capable and relatively secure. This balance is essential for integrating AI into a professional office environment where data protection is a priority.
The role of the human user
While the webinar shows how to make Claude more autonomous, it also emphasizes the importance of the human user in the process. The AI is a tool that needs to be directed, styled, and supervised. The training shows how to set the rules and provide the context that the AI needs to be successful.
The user acts as the architect of the assistant, deciding which files to include and which styles to prioritize. By setting up 'Projects' and 'Styles,' the human provides the framework within which the AI operates. This means the quality of the AI's output is directly tied to the quality of the instructions and context provided by the user.
The webinar teaches users how to be better 'managers' of their AI assistants. Instead of just asking for one-off answers, they learn how to build a system that understands their work. This shift in mindset is what allows users to move from basic chat interactions to a more sophisticated, integrated workflow that actually improves their daily productivity.
Webinar logistics and details
The webinar is a structured learning event designed to provide a comprehensive overview of Claude's capabilities. It is scheduled for December 8, 2026, and will run for two and a half hours. The cost of the session is 69 Euro, which covers the guided instruction and the practical examples provided by the c't editor.
Participants do not need any programming knowledge to attend the session. The focus is on practical usage for everyday work tasks, making it suitable for a wide range of professionals. The only requirements are a current web browser and a sense of curiosity about how to improve your work with AI.
To get the most out of the session, it is highly recommended that users have a free Claude account ready. This will allow them to follow along with the steps shown by Bager in real-time. After the webinar, they will be able to apply these techniques to their own accounts to start building their own custom assistants.
Summary of the transition
The webinar represents a shift from using AI as a search engine to using it as a specialized assistant. Most users start with simple prompts, but the training shows how to move toward a much more integrated experience. This involves setting up specific workspaces, teaching the AI a personal style, and providing it with the necessary knowledge files.
By learning these techniques, users can automate multi-step tasks and handle complex office workflows more efficiently. The training also ensures that this transition is done safely by covering security risks and identifying the AI's limitations. It provides a complete picture of how to move from a first chat to a fully functional, personalized assistant.
Ultimately, the webinar aims to make users smarter by showing them how to leverage the full potential of Claude. It moves the conversation from 'what can AI do?' to 'how can I make AI do this specific task for me?' This practical approach is what helps people integrate AI into their daily lives and work routines effectively.
Final thoughts on AI usage
As AI tools become more common in the workplace, the ability to use them effectively will become a key skill. The webinar from c't provides a clear path for those who want to move beyond the basics. It focuses on the practicalities of setting up, styling, and managing an AI assistant to handle real-world tasks.
By focusing on context, files, and rules, the training helps users create a more reliable and personalized experience. It shows how to turn a generic chatbot into a tool that understands your specific writing tone and your unique project data. This is the key to moving from simple questions to complex, multi-step workflows.
The session is designed to be accessible to everyone, regardless of their technical background. It provides a structured way to explore the power of Claude while keeping security and accuracy in mind. For anyone looking to improve their daily work, it offers a practical guide to making AI a truly useful part of their professional life.
Key takeaways for attendees
Attendees of the webinar will walk away with a clear understanding of how to organize their AI interactions. They will learn how to use 'Projects' to keep different work areas separate and how to use 'Styles' to maintain a consistent voice. These are the building blocks for a clean and organized AI workspace.
They will also see how to use 'knowledge files' to ground the AI's responses in specific data. This is a major step up from general knowledge, as it allows the AI to work with the user's own documents and folders. The training will show how to do this effectively while keeping the sidebar and overall interface manageable.
Finally, the webinar will cover the 'Cowork' function for multi-step tasks and the necessary security settings to keep data safe. By the end of the session, participants will have a roadmap for building their own assistants. They will know how to give instructions, check the results, and save recurring workflows as skills for future use.
Practical steps for the reader
If you are interested in these techniques, you can start by creating a free Claude account today. This will allow you to explore the basic interface and begin experimenting with the features mentioned in the webinar.
Sources and links
- Reportingheise.de: the article we sourced
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.
Listen to the AI Wire headlines
Listen to the AI WireListen to our daily AI shows every morning: on your commute, in the shower, wherever you are
Listen to the daily showsFirst reported by heise.de · Read the original. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.
