🧠 The rag is there: the AI avatars finally rely on your documents

On Complement, a powerful new feature has just appeared: RAG can now be activated for all your courses .

It's a feature that's still relatively unknown, but it can make all the difference for certain types of content. Let us explain.

🔍 What exactly is RAG?

RAG stands for Retrieval Augmented Generation . Simply put, this allows the AI ​​avatar to formulate its answers based on the documents you've added to your course. Rather than inventing a generic answer, the AI ​​will first search your own content : PDFs, texts, industry resources, etc.

And as always on Complement, transparency is key : the AI ​​indicates the source used to build its answer, so that the learner can delve deeper into the subject if they wish.

🧪 How does it work in Complement?

When you import additional documents into your module (PDF support, reference text, link to a reliable site…), they are indexed.

If you choose to enable RAG , then these resources become the reference basis used by the AI ​​to answer student questions.

A simple checkbox in the creation interface… and it's in place.

The learner can still consult the documents independently, but with the RAG, they also benefit from an AI tutor capable of drawing contextualized and precise answers from them .

✅ When is it useful?

RAG becomes particularly interesting in certain cases:

When you are dealing with a specific, technical or sensitive topic , when there are obscure acronyms, when internet searches do not provide a consensus on a subject: when it is important that the AI's answers are 100% aligned with your content and not with what is said online.

For example, it is ideal for internal training (security, procedures, company policy) or specialized content (regulatory, financial, legal, health, etc.).

⚠️ And when should it not be activated?

For more general or theoretical training, or if you don't have documents to provide, AI works very well without RAG. The avatar will provide a smooth, quick, and often sufficient response for simple questions.

Enabling RAG in these cases may slightly increase response times, without any real benefit.

🤔 Are you hesitating to activate it?

If you're unsure about the suitability of RAG for a particular module, the best approach is to test it . Our team can guide you on best practices and demonstrate the difference live.

RAG is a superpower , but like all superpowers: it activates at the right time.

✨ Want to learn more about Complement?

👉 Try it for free or 👉 Request a demo

‍