Skip to content

Getting Started

Hello and welcome to this Getting Started guide. We do not assume anything other than a very fundamental knowledge of python.

Beginner's Guide to Indexify: Building a Wikipedia Information Retrieval System


In this guide, we'll walk you through creating an online ingestion pipeline for Wikipedia pages. This pipeline will demonstrate how to:

  1. Extract structured information (Named Entity Recognition) from web pages using Large Language Models (LLMs)
  2. Break down (chunk) the text, create embeddings, and store them in a vector database (LanceDB in this example)

By the end of this tutorial, you'll be able to:

  1. Use advanced LLMs like GPT-4 or Mistral to answer questions based on indexed information (Retrieval-Augmented Generation or RAG)
  2. Retrieve the Named Entities extracted from the text
  3. Use a User Interface to visually debug your pipelines and inspect how pages are broken down into chunks

Let's get started!


Before we begin, make sure you have:

  1. Python 3.7 or higher installed
  2. Basic knowledge of Python programming
  3. Familiarity with command-line interfaces
  4. An OpenAI API key (for using GPT models)


You'll need three separate terminal windows open for this tutorial:

  1. Terminal 1: For downloading and running the Indexify Server
  2. Terminal 2: For running Indexify extractors (handling structured extraction, chunking, and embedding)
  3. Terminal 3: For running Python scripts to load and query data from the Indexify server

We'll use the following notation to indicate which terminal to use:

( Terminal X ) Description of Command
<command goes here>

Understanding Indexify Components

Before we dive in, let's briefly explain the key components of Indexify:

  1. Indexify Server: The central coordinator and data ingestion API.
  2. Extractors: Specialized workers designed to perform specific data processing tasks (e.g., embedding data, generating summaries, or extracting features from unstructured data).
  3. Extraction Graph: A declarative YAML file that chains together extractors into a complex pipeline.

Also before we look into creating pipelines for ingestion and query, it is best to lay out the directory structure of our project.

Directory Structure
├── venv/                  # Virtual environment (created by python3 -m venv venv)
├── graph.yaml             # Extraction graph definition
├──               # Script to create the extraction graph
├──              # Script to ingest Wikipedia data
├──               # Script to query the indexed data
└── indexify               # Indexify server executable (downloaded by curl command)

To use this structure:

  1. Create a new directory called indexify-tutorial.
  2. Navigate into this directory in your terminal.
  3. Create the virtual environment and activate it (discussed below).
  4. Create each of the .py and .yaml files in the root of this directory (discussed below).
  5. Run the curl command to download the Indexify executable into this directory (discussed below).

This structure keeps all the components of our tutorial project organized in one place, making it easy to manage and run the different scripts.

Step 1: Setting Up the Indexify Server

To start the indexify server, we have to open up a terminal and put in the download command.

Indexify Terminal 1

Let's start by downloading and running the Indexify server:

( Terminal 1 ) Download Indexify Server
curl | sh
./indexify server -d

This command creates two important endpoints:

  1. Ingestion API: http://localhost:8900
  2. User Interface: http://localhost:8900/ui

The Ingestion API is used for uploading content and retrieving data from indexes and SQL tables, while the User Interface provides a dashboard for visualizing extraction graphs, content, and indexes.

Step 2: Creating a Virtual Environment

It's good practice to use a virtual environment for Python projects. Let's create one and install the necessary packages:

( Terminal 2 ) Install Dependencies
python3 -m venv venv
source venv/bin/activate
pip3 install indexify-extractor-sdk indexify wikipedia openai langchain_community

Step 3: Setting Up Indexify Extractors

The next step is of course setting up extractors to extract data reliably at scale. Extractors are the very soul of the pipelines at Indexify. Extractors are used for structured extraction from un-structured data of any modality. For example, line items of an invoice as JSON, objects in a video, embedding of text in a PDF, etc.

For the purpose of this tutorial we will be using extractors that have already been created. You can find the full list here

Indexify Terminal 2

Extractors consume Content which contains raw bytes of unstructured data, and they produce a list of Content and features from them.


If you want to read and understand how to build a custom extractor for your own use case, go through the following section. However, if you want to use a built-in available extractor jump to the next section.

Using Custom Extractors

An extractor in Indexify is designed to process unstructured data. It receives data in a Content object and transforms it into one or more Content objects, optionally adding Feature objects during extraction. For example, you could split a PDF into multiple content pieces, each with its text and corresponding embedding or other metadata.

Key Concepts

  1. Content: Represents unstructured data with properties:

    • data: Raw bytes of the unstructured data
    • content_type: MIME type of the data (e.g., text/plain, image/png)
    • Feature: Optional associated feature (embedding or JSON metadata)
  2. Feature: Extracted information from unstructured data, such as embeddings or JSON metadata.

Building a Custom Extractor

Building a custom extractor is an easy 4 step process. We will walk through this while laying down each step in a sequential manner. To read a full guide on building custom extractors, read the official documentation for developing extractors.

Step 1: Clone from Template

Use the following command to create a template for your new extractor:

( Terminal 2 ) Download Code Template
curl | tar -xz  indexify-extractor-template-main
Step 2: Implement the Extractor

In the template, you'll find a MyExtractor class in the file. Implement the extract method, which takes a Content object and returns a list of Content objects.

Basically extract method takes a Content object which have the bytes of unstructured data and the mime-type. You can pass a list of JSON, text, video, audio and documents into the extract method. It then returns a list of transformed or derived content, or a list of features.

For instance, in the following example snippet we iterate over a list of content, chunk each content, run a NER model and an embedding model over each chunk and return them as features along with the chunks of text.
def extract(self, content: Content) -> List[Content]:
    Extracts features from content.
    output: List[Content] = []
    chunks = content.chunk()
    for chunk in chunks:
        embedding = get_embedding(chunk)
        entities = run_ner_model(chunk)
        embed_chunk = Content.from_text(text=chunk, feature=Feature.embedding(name="text_embedding", values=embedding))
        metadata_chunk = Content.from_text(text=chunk, feature=Feature.metadata(name="metadata", json.dumps(entities))),
        output.append([embed_chunk, metadata_chunk])
    return output
Step 3: Define Extractor Properties

Add the following properties to your extractor class:

  • name: The name of your extractor
  • description: A detailed description of what your extractor does
  • system_dependencies: List of dependencies for packaging in a Docker container
  • input_mime_types: List of input data types your extractor can handle (default is ["text/plain"])
Step 4: List Dependencies

Create a requirements.txt file in your extractor's folder to list any Python dependencies.

Deploying Locally

Local Installation

Install your extractor locally to make it available to the Indexify server:

indexify-extractor install-local custom_extractor:MyExtractor 

Joining with Control Plane

Connect your extractor to the Indexify server to receive content streams:

indexify-extractor join-server

By following these short steps, you can create, and locally deploy custom extractors in Indexify, allowing you to integrate specialized data processing capabilities into your Indexify pipelines.

Using Available Extractors

As mentioned before, for the purpose of this tutorial, we already have Extractors written, deployed and tested. Now, let's download three essential extractors:

( Terminal 2 ) Download Indexify Extractors
source venv/bin/activate
indexify-extractor download tensorlake/openai
indexify-extractor download tensorlake/minilm-l6
indexify-extractor download tensorlake/chunk-extractor
indexify-extractor join-server

Don't forget to set your OpenAI API key:

export OPENAI_API_KEY=your_api_key_here

Now, let's start all available extractors:

( Terminal 2 ) Starting Extractor Workers
indexify-extractor join-server

Step 4: Defining Our Data Pipeline

We’ll define our data pipeline using a YAML file to process text documents by splitting them into chunks, extracting entities, and embedding the chunks in parallel. The following diagram outlines the Indexify end-to-end pipeline.

Extraction Policy Graph

Let us create (or open) a file named graph.yaml with the following content:

name: "wiki_extraction_pipeline"
  - extractor: "tensorlake/openai"
    name: "entity-extractor" 
      system_prompt: "Extract entities from text, and return the output in JSON format."
  - extractor: "tensorlake/chunk-extractor"
    name: "chunker"
      chunk_size: 1000
      overlap: 100
  - extractor: "tensorlake/minilm-l6"
    name: "wikiembedding"
    content_source: "chunker"

This YAML file defines three extraction policies: 1. entity-extractor: Extracts named entities from the text 2. chunker: Splits the text into smaller chunks 3. wikiembedding: Creates embeddings for the chunks

These are the three key extractors are utilized to process and analyze the input content from Wikipedia:

  1. OpenAI Extractor (tensorlake/openai): Uses OpenAI’s language models for entity extraction from various input types (text, PDF, images). The extractor can be customized with specific system and user prompts to tailor its output.

  2. Chunk Extractor (tensorlake/chunk-extractor): Breaks down text into chunks, offering flexibility in chunk size and overlap.

  3. MiniLM-L6 Extractor (tensorlake/minilm-l6): Generates embeddings of the text chunks for semantic search and retrieval.

These extractors work in concert to transform raw, unstructured input into processed, indexed, and easily retrievable information, forming the backbone of the Indexify pipeline for tasks such as entity recognition, text segmentation, and semantic embedding. You can learn more about different types of available extractors and their usage here.

Step 5: Creating the Extraction Graph

Now, let's create a Python script to set up our extraction graph using the YAML file we just created.

Create a file named with the following content:
from indexify import IndexifyClient, ExtractionGraph

client = IndexifyClient()

def create_extraction_graph():
    extraction_graph = ExtractionGraph.from_yaml_file("graph.yaml")

if __name__ == "__main__":

Run this script to create the extraction graph:

( Terminal 3) Create Extraction Graph
source venv/bin/activate
python3 ./

Step 6: Loading Data

Now that we have our extraction graph set up, let's create (or open) a script to load the Wikipedia data into our pipeline.

Indexify Terminal 3

Create a file named with the following content:
from indexify import IndexifyClient, ExtractionGraph
from langchain_community.document_loaders import WikipediaLoader

client = IndexifyClient()

def load_data(player):
    docs = WikipediaLoader(query=player, load_max_docs=1).load()

    for doc in docs:
        client.add_documents("wiki_extraction_pipeline", doc.page_content)

if __name__ == "__main__":
    load_data("Kevin Durant")
    load_data("Stephen Curry")

Run this script to ingest data into Indexify:

( Terminal 3) Ingest Data
source venv/bin/activate
python3 ./

Step 7: Querying Indexify

Now that we have data in our system, let's create a script to query Indexify and retrieve information.

You can query Indexify to -

  1. List ingested content by extraction graph. You can also list content per extraction policy.
  2. Get extracted data from any of the extraction policies of an Extraction Graph.
  3. Perform semantic search on vector indexes populated by embedding extractors.
  4. Run SQL Queries on structured data(not in this tutorial).

For now let us create a file named with the following content:
from indexify import IndexifyClient
from openai import OpenAI

client = IndexifyClient()
client_openai = OpenAI()

# Get entities
ingested_content_list = client.list_content("wiki_extraction_pipeline")
content_id = ingested_content_list[0].id
entities = client.get_extracted_content(

# Get chunks
chunks = client.get_extracted_content(

def query_database(question: str, index: str, top_k=3):
    retrieved_results = client.search_index(name=index, query=question, top_k=top_k)
    context = "\n-".join([item["text"] for item in retrieved_results])
    response =
                "role": "user",
                "content": f"Answer the question, based on the context.\n question: {question} \n context: {context}",
    return response.choices[0].message.content

if __name__ == "__main__":
    index_name = "wiki_extraction_pipeline.wikiembedding.embedding"
    indexes = client.indexes()
    print(f"Vector indexes present: {indexes}, querying index: {index_name}")
            "What accomplishments did Kevin durant achieve during his career?",

Run this script to query the indexed data:

( Terminal 3 ) Run our RAG query
OPENAI_API_KEY=your_api_key_here python3 ./

You should see a response summarizing Kevin Durant's career accomplishments based on the indexed Wikipedia data.

During his career, Kevin Durant has achieved numerous accomplishments, including winning two NBA championships, an NBA Most Valuable Player Award, two Finals MVP Awards, two NBA All-Star Game Most Valuable Player Awards, four NBA scoring titles, the NBA Rookie of the Year Award, and being named to ten All-NBA teams (including six First Teams). He has also been selected as an NBA All-Star 14 times and was named to the NBA 75th Anniversary Team in 2021. Additionally, Durant has won three gold medals in the Olympics as a member of the U.S. men's national team and gold at the 2010 FIBA World Championship


Congratulations! You've successfully set up an Indexify pipeline for ingesting, processing, and querying Wikipedia data. This beginner-friendly guide has walked you through:

  1. Setting up the Indexify server and extractors
  2. Defining an extraction graph for processing Wikipedia pages
  3. Ingesting data into the system
  4. Querying the processed data using semantic search and GPT-3.5

Indexify's fault-tolerant design ensures reliability and scalability, making it suitable for mission-critical applications. You can now explore more advanced topics and integrations to further enhance your information retrieval and processing capabilities.

Next Steps

To continue your journey with Indexify, consider exploring the following topics in order:

Topics Subtopics
Intermediate Use Case: Unstructured Data Extraction from a Tax PDF - Understanding the challenge of tax document processing
- Setting up an Indexify pipeline for PDF extraction
- Implementing extractors for key tax information
- Querying and retrieving processed tax data
Key Concepts of Indexify - Extractors
• Transformation
• Structured Data Extraction
• Embedding Extraction
• Combined Transformation, Embedding, and Metadata Extraction
- Namespaces
- Content
- Extraction Graphs
- Vector Index and Retrieval APIs
- Structured Data Tables
Architecture of Indexify - Indexify Server
• Coordinator
• Ingestion Server
- Extractors
- Deployment Layouts
• Local Mode
• Production Mode
Building a Custom Extractor for Your Use Case - Understanding the Extractor SDK
- Designing your extractor's functionality
- Implementing the extractor class
- Testing and debugging your custom extractor
- Integrating the custom extractor into your Indexify pipeline
Examples and Use Cases - Document processing and analysis
- Image and video content extraction
- Audio transcription and analysis
- Multi-modal data processing
- Large-scale data ingestion and retrieval systems

Each section builds upon the previous ones, providing a logical progression from practical application to deeper technical understanding and finally to customization and real-world examples.

Happy coding!