🔌 RAG Assistant for EV Charging Support

About this simulation

Retrieval-Augmented Generation (RAG) pairs a large language model with a searchable knowledge base so answers are grounded in real documents instead of the model's memory alone. This scene visualises that pipeline for an EV charging support desk: manuals, fault-code tables and FAQ entries live as points scattered through a 3D embedding space, clustered by topic.

How it works

• Embedding — the incoming question is converted into a point (vector) in the same space as the documents, landing near the cluster it's semantically closest to.
• Vector search — the assistant measures distance from the query point to every document point; closer points are more semantically similar.
• Retrieval — the top-K nearest chunks above the similarity threshold light up and are pulled toward the query.
• Generation — the retrieved chunks travel into the language-model core, which composes a grounded answer shown on the terminal panel.

Controls

• Support question — selects a sample EV charging query, each targeting a different topic cluster.
• Chunks to retrieve (top-K) — how many nearest document chunks are pulled into the context window.
• Similarity threshold — the minimum closeness a chunk needs to qualify; raise it to see fewer, tighter matches.
• Knowledge base size — how many document chunks populate the vector space.
• ▶ Ask the assistant — runs the embed → search → retrieve → generate sequence. Drag to rotate, scroll to zoom.

Did you know?

RAG became popular because fine-tuning a model on every new fault code or price change is slow and expensive — updating a vector database with a fresh document is almost instant, and the model always cites current, retrievable sources.