LlamaIndexllama-index-core 0.14 · Python 3.10+
Dashboard
0%
1
Curious builder0 XP earned · 300 to level 2
0 daysFinish a lesson to begin
Badge collection0 of 6 unlocked
30 small wins to finish your pathNext lesson →

LlamaIndex logoLlamaIndex overview

LlamaIndex is a data framework that connects a language model to your own documents, so it can answer from them instead of guessing.

Last updated: 28 Sep, 2026 · LlamaIndex 0.14

Jerry Liu started the project in late 2022 as GPT Index, and it was renamed LlamaIndex as it grew into a general framework. It is released under the MIT licence, and this course uses llama-index-core 0.14.25. The company behind it also offers hosted services: LlamaParse for document parsing and LlamaCloud for managed indexes.

This course builds one thing across every lesson: a help-centre assistant for a small shop. By the last lesson it answers a customer's question from the shop's own help pages, cites the page it used, checks who is asking, and refuses when the pages do not hold the answer.

Readers, node parsers, indexes and retrievers

The framework is a small core package plus hundreds of integration packages, one per model provider, embedding model or vector store. Its main pieces are readers that load documents, node parsers that split them into chunks, indexes that store the chunks, retrievers that find the chunks closest to a question, and query engines that turn retrieved chunks into an answer.

Retrieval-augmented generation, or RAG, is the common way to make a model answer from a company's own documents: find the passages that match the question, put them in the prompt, and have the model answer from them. LlamaIndex covers every step of that flow, and this course walks through it in order.

One question through the assistant
Load and splitinto chunksEmbedchunks and the questionRetrieveclosest, filtered by askerAnswerciting the sourceMeasurelabelled questions
Hover or tap a piece to see what it is and which lesson built it.
Trace a question

Pick one to watch it run, step by step.

Everything in this course, and where it is going
Retrieval from first principlesLoading and chunkingQueryingBetter retrievalKeeping it runningwhy retrievalembeddingsa first indexreading a folderchunk sizemetadataretrieversanswers with sourcespermissionskeyword searchhybrid searchmeasuring itsaving an indexupdating documentsthe help-centre assistant
Hover or tap a piece to see what it is and which lesson built it.

all-MiniLM-L6-v2 and a stand-in answerer

Embeddings come from all-MiniLM-L6-v2, a small open model that runs on your own computer, so every retrieval result in the course is a real one. Answers come from a small stand-in model you write later in the course, and one lesson prints the exact prompt a real model would receive. That way nothing here needs a paid key, yet every score and every retrieved chunk is genuine.

Python, disk space and prerequisites

  • Python 3.10 or later, and about 1 GB of disk space for PyTorch and two small models.
  • The Python for AI course, and the LLM Fundamentals course for tokens and prompts, are enough background.
Keyless from start to finish
Every example in this course runs with no API key. The embedding model downloads once and then loads from your disk, and the answering model is a stand-in you can read line by line.

Where a help centre like this fits

  • A support assistant that answers from policy pages and links the page it quoted.
  • An internal search over runbooks or a wiki, where staff need the exact passage, not a guess.
  • Any question box over documents that change often, where re-reading everything per question is too slow.
Back toAll frameworks

Every expert started right here.