The Expanded eLeDia.ai | Tutor Stack
The eLeDia.ai Tutor is not a single plugin, but rather a coordinated system. Each component performs a clearly defined task: one displays the chat, the next provides the knowledge, the third formulates the answer, and the fourth securely integrates the AI into Moodle. Together, they form an AI tutor that responds using your actual course content—verified, accurate, and within your own environment.
Stack Overview
The easiest way to explain it is in two steps: First, content flows in (preparation), and then a question flows through (operation).
1. Preparation: Building Knowledge
RagIngest monitors the courses you have approved. When content is created or modified, it extracts the text and sends it to LiteRAG. There, the content is stored as searchable chunks in the Moodle database. This creates—in a controlled, opt-in manner—the knowledge base from which the tutor will later draw when responding.
2. Operation: Answer a question
When a learner asks a question in the tutor block, it is processed on the server side by LiteRAG. LiteRAG searches the knowledge base, filters the results based on that specific learner’s Moodle permissions, and uses a language model to generate a response with sources. If necessary, LiteRAG retrieves additional live data from Moodle via eLeDia MCP —such as deadlines, grades, or the learner’s progress. The finished response is sent back to the chat.
You don't have to implement everything at once. The system scales from a simple chat to a fully RAG-based tutor.
Level 1: Pure LLM chat. Just the tutor block. It runs independently, without any additional plugins: an AI chat in Moodle that responds using the language model. Ideal for a quick start.
Level 2: RAG-based tutor. The tutor module plus RagIngest, LiteRAG, and eLeDia MCP. Now the tutor responds based on your actual course content, cites sources for its answers, and, if desired, takes the student’s individual learning progress into account.
Why four plugins?
This clear separation is a deliberate design choice that offers specific advantages:
Security through accountability. The block displays only the user interface and keeps all sensitive data on the server. LiteRAG filters every request against Moodle permissions. eLeDia.ai | MCP enforces the logged-in user’s permissions for every data access. No component can do more than it is allowed to.
Control over knowledge. Because RagIngest separates the input from the response, you can decide precisely—course by course—exactly what the tutor is allowed to know.
Interchangeability. LiteRAG is the included backend, but the tutor block can, in principle, also interface with a different RAG/tutor backend. The language model is OpenAI-compatible and can therefore be freely selected.
Data sovereignty. All four components run within your Moodle environment. Your content, your infrastructure, your model.
LiteRAG vs. RAGServer
LiteRAG for a quick and easy start.
LiteRAG is the streamlined version built into Moodle. It runs as a Moodle plugin directly within your installation, without any additional infrastructure: no external vector database, no embeddings—just classic full-text database search. It receives its data via opt-in through RagIngest—specifically, only the courses you choose to share. This makes it ideal for a quick start, small- to medium-sized setups, and administrators who want to minimize ongoing effort. LiteRAG is therefore available as an open-source plugin.
RAGServer for More Complex Requirements
RAGServer is the powerful, standalone version. It runs as a separate Docker service alongside Moodle, scans an entire Moodle installation (rather than just selected, shared courses), and can also integrate additional, external knowledge sources. It is therefore intended for large installations and more demanding use cases where LiteRAG’s built-in full-text search and limited corpus are no longer sufficient.
Learn about the AI Tutor for your Moodle system
Would you like to add an AI-powered tutor to Moodle and give learners a new way to access your course content?
We would be happy to advise you on integrating the eLeDia.ai tutor, setting up the appropriate infrastructure, and selecting a suitable language model.
Your trust, our responsibility
eLeDia is certified according to ISO/IEC 27001. This ensures that all areas - from hosting and individual development to support - meet the highest security standards. We guarantee the protection of sensitive data through external audits.