AI connected to learning and practice
The LSyM assistant brings campus information, simulator activity and the capabilities of different AI models into one conversation. Instructors can investigate, interpret and support learning without leaving their workspace.
Ask, verify and decide with the right context.
An assistant that understands the environment
The assistant is part of the instructor panel and works with the context already present in the system: courses, content, participants, progress, exercises and simulation activity.
In local mode, requests use read-only MCP tools. The model receives only information the user is allowed to consult, and the interface shows which tools contributed to the answer.
If one source is unavailable, the assistant can continue with the other: ILIAS and LSyM keep their responsibilities separate.
Locate courses, content, practices and materials within the authorised context.
Review participants, progress, exercises, runs and activity according to instructor permissions.
Summarise information, explain results and keep follow-up questions in the same conversation.
Course materials
“Consult course materials” adds a verification layer. The instructor selects an accessible course and the answer links its claims to extracts from pages, modules and text files.
They are checked again before a saved answer is shown. If a resource is no longer authorised, it is removed from the visible context. If there is not enough evidence, the assistant says so instead of completing the answer with unsupported information.
The current version uses verifiable text retrieval. Semantic search with embeddings and extraction from PDF, DOCX and multimedia content belong to the next stage.
Two AI modes
LM Studio can consult ILIAS and LSyM through read-only MCP integrations. It is the appropriate mode for questions about courses, activity, progress, exercises and authorised materials. Tools used are shown with the response, and technical credentials never reach the browser.
OpenAI, Google Gemini, Anthropic Claude, DeepSeek and Groq can be enabled for general training explanations. They do not receive course documents or internal LMS and simulator tools, and the interface warns users not to enter personal or confidential information.
Each provider keeps a separate conversation. Switching models does not mix histories or contexts; the instructor can clear the conversation and start again.
Also inside the simulator
The consultation experience is also available in the simulator's native panel, with a provider selector, separate history and asynchronous requests so the interface remains responsive while an answer is prepared.
The LMS keeps the academic context and the simulator supplies the operational context. Both remain the sources of truth for their data; AI helps interpret them but does not replace controls or authorised workflows.
Human and technical control
Access is resolved through LMS permissions, not direct database queries.
Changes, launches and other actions still use authenticated ILIAS and LSyM workflows.
Tokens and keys remain on the server and are not added to HTML or visible conversations.
Conversations are isolated by user, panel and provider and can be deleted from the panel.
The assistant follows the active language; the interface is available in Spanish and English.
Available now
The value of AI is not answering for its own sake, but finding the right evidence, explaining it and turning it into a reviewable training decision.