Rethinking the LMS in an AI Age: Why the “Digital Filing Cabinet” is Dead
In 2022, in the middle of the creative re-thinking university teaching when COVID-19 has just been breaking up cemented habits, I reached a breaking point. I was managing courses with over 300 individual items—scripts, videos, exercises, examples, whiteboard links, and external links. In my university’s standard LMS, **ILIAS**, this meant a labyrinth of folders, sub-folders, and sub-sub-folders.
The result? Total friction. It felt like every important resource was at least “10 clicks away.”, spending almost half an hour just for uploading and linking the resources for a single day. Despite my best efforts to organize, my inbox was flooded with students asking for materials that were already there. They constantly complained about a “lack of structure,” and honestly, I couldn’t blame them. I was spending more time acting as a digital librarian than as an educator.
Then, a student said something that stuck with me. Looking at the interface of our university systems, he remarked: “Everything here feels so *old*.”
He wasn’t just talking about the buildings or the furniture. He was talking about the digital experience. Outside the lecture hall, these students live in a world of sleek, powerful tools like ChatGPT and Gemini. They are used to instant answers and intuitive interfaces. When they step into the “old world” of traditional university systems, the disconnect is jarring.
This was the spark for the Learning Companion, an abundance of ideas that has evolved into a modern learning stack with the work of several students. We realized that if we want to stay relevant in the AI age, we have to stop “managing learning” and start providing a modern learning environment.
1. From “10 Clicks Away” to Contextual Embedding and views
The biggest lesson from the 300-item nightmare was that structure shouldn’t be a hierarchy of folders; it should be contextual view**.
Instead of forcing students to leave their workspace to find a resource in a siloed LMS, the Learning Companion brings the material *to them*. Views allow for recency views, timeline-based views, topic-based views, and structure-based views, as well as a conversational interface. The material views can also be embedded inside Miro boards, which are used throughout the course, from lectures via workshops to students’ own portfolio projects. Because creative and collaborative work happens on the whiteboard, the Companion allows us to embed curated views—videos, texts, and prompts—directly where the students are working. No more clicking through folders; the support is right there, in context.
2. Moving from Content Provision to Support Services
In the age of AI, information is no longer a scarce resource. Students are already using LLMs to summarize lectures and explain concepts. If we just provide a “digital filing cabinet” like ILIAS, we are essentially providing a service that ChatGPT or Gemini can already do better.
The Learning Companion shifts the focus to **curated support services**. It blends (and mutually enhances):
- Human Expertise and Personal Connection: High-level pedagogical guidance and mentoring, emotional connection
- Agentic AI Support: An LLM-based chatbot that isn’t just a generic assistant. It is “course-aware,” trained on the specific transcripts, web resources, and materials provided by the teacher. It provides 24/7, course-specific support that helps students navigate the *abundance* of information rather than just drowning in it.
3. Online-First, but Hybrid-Ready
Traditional systems often feel like digital versions of a physical classroom—rigid and linear. The Learning Companion is designed **online-first**, built on a cloud-native, modern technology stack (headless CMS, stateless containers).
It is optimized for both synchronous and asynchronous learning. Whether a student is sitting in the front row of a lecture hall or working from a café at midnight, they have the same powerful, lightweight interface. By using “convention over configuration,” we’ve made it easy for teachers to set up and maintain these environments without getting bogged down in administrative menus.
4. Assessing Competence, Not Recall
The “everything feels so old” sentiment hits hardest during assessments. If an exam only tests for the recall of facts, it’s an exam for a world that no longer exists.
The Learning Companion is built to support a new era of assessment. We are moving away from prioritizing what a student can memorize and toward their **competence in conversing with an AI to solve a task**. This includes:
- Problem Formulation: Can you “prompt” an AI to help solve a complex modeling problem?
- Critical Interpretation: Can you spot the errors or biases in the AI’s output? Can you learn from the steps of the AI?
- Iterative Solving: Can you use the Companion’s tools to refine a solution? Can you help the AI to make the next steps?
The Future is a Companion, Not a System
We built the Learning Companion because we wanted to bridge the gap between the “old world” of university administration and the “modern world” of AI-driven work. We wanted to turn the “10 clicks of frustration” into a single point of intelligent support.
In an age of abundance, the goal of education isn’t to store content—it’s to provide the scaffolding that helps students climb.
View more information at https://learningcompanion.de