Qdrant, a leading vector database and vector similarity search engine, and Lyzr, an enterprise agent framework known for its simplicity and data privacy, have partnered to launch the ‘Private Perplexity’ application for enterprise customers. With the ‘Knowledge Search’ app, enterprises can deploy a 100% private and secure document and content search engine for their organization in their cloud account.

How to access the demo app?

Visit https://qdrant.lyzr.ai/ to test the micro-app built on Lyzr and Qdrant. Lyzr Qdrant

Key Features

  • You can add any number of source data at any point in time. The vectorization and indexing are dynamic and happen as and when new data is added.
  • When you search, you also get the data citations that will be useful for locating the source file.
  • You can save, share, and even bookmark the search results
  • The micro-app is open-sourced. You can modify the UI, add your logo and use the app as is for your organization.

Looking to deploy the solution using open-source versions?

Here is the colab notebook to help you get started fast.

Let’s see how it works.

  1. Import the necessary libraries. OpenAI GPT4 Turbo is the LLM, Lyzr is the agent framework and Qdrant is the vector store.
import openai
import qdrant_client
from lyzr import SearchAgent
  1. Pass the OpenAI and Qdrant credentials
os.environ["OPENAI_API_KEY"] = ""
openai.apikey = os.environ["OPENAI_API_KEY"]
  1. Setup the Qdrant vector store with Lyzr Search Agent
client = qdrant_client.QdrantClient(
    api_key= QDRANT_CLIENT_URL,

vector_store_params = {

search = SearchAgent()
  1. Pass the data that the RAG pipeline will access. In this example, we pass a PDF document as the data source.
agent = search.add_pdf(input_files=["pdf_file_path"],vector_store_params=vector_store_params)
  1. Ask your search query.
response = agent.query("Ask your query")

That’s it. You have built a personal, fully-private, open-source version of the perplexity-style search engine for your organization in just 5 steps.