Vectara Beginner App Tutorial: Showcase The Creation Of Vectara App In Legal Use Case

Friday, November 03, 2023 by sanchayt743
Vectara Beginner App Tutorial: Showcase The Creation Of Vectara App In Legal Use Case

1. Introduction to the Vectara Ecosystem: 🌌

Welcome to the realm of Vectara, a platform where search transcends to new horizons powered by Generative AI. Here, I will unravel the essence of the Vectara ecosystem, its fundamental workings, and provide a visual journey through Vectara's official materials to enhance your comprehension.

πŸš€ Overview of Vectara Ecosystem

Vectara is on a mission to redefine how users interact with data and knowledge, facilitating a seamless journey from a user query to the most relevant response. The platform harbors a complete, but easy to customize, search and summarization pipeline, making it an API-based easy-to-use platform to build applications powered by semantic search and Generative AI. Through Vectara, developers are empowered to create GenAI applications with a state-of-the-art retrieval engine and summarization capability, thus elevating the user experience to a realm where questions meet precise answers.

πŸ›  Fundamental Workings and Workflow

The heartbeat of Vectara is its pure neural search platform enriched with production-ready natural language processing. The workflow is simple, yet powerful:

  1. Data Ingestion: Ingest your data into Vectara's corpus using the Indexing API.
  2. Query Execution: Utilize the Search API to run queries against the indexed data, retrieving highly relevant information swiftly.

The beauty of Vectara lies in its API-addressable platform, which is a canvas for developers to paint their GenAI solutions, embedding them within their applications.

🎨 Dive into Vectara's Console

To truly grasp the potential of Vectara, let's delve into its console which is the epicenter of managing your account:

  • Creating Corpora: Jumpstart your journey by establishing a corpus, your data's secure haven, ready for querying. Here’s how to navigate this straightforward process:

    • Name Your Corpus: Give your corpus a unique identifier.
    • Provide a Description: Briefly describe the purpose or content of your corpus.
    • Choose an Embedding Model: Select the embedding model that best aligns with your needs.
    • Specify Filter Attributes (Optional): You have the option to add filter attributes for additional refinement.

    Just like that, your corpus is configured and ready to receive data!

  • API Access Management: Vectara empowers you with the tools to seamlessly manage API access. Utilize the API access tab, visible in the sidebar once the requisite permissions are granted, to effortlessly create and manage your API keys and app clients. This is your gateway to leveraging Vectara’s robust search capabilities.

  • Team Collaboration: Enhance your project by inviting team members to the Vectara console. Assign roles, establish permissions, and cultivate a collaborative environment, all aimed at refining and perfecting your search solutions.

  • Billing Management: Maintain oversight of your account usage and manage your billing details to ensure uninterrupted access to Vectara’s services. Navigate the ecosystem with ease, secure in the knowledge that your account is in good standing.

In this section, we've skimmed the surface of Vectara's offerings. As we delve deeper into our chosen use case in the next section, the utility and power of Vectara will unfold further, painting a clearer picture of how it can be harnessed for Customer Support applications.

Embark on an insightful journey through this tutorial, where we unveil the Legal Consultation Application meticulously crafted using Streamlit, Vectara, and Langchain. This innovative application is engineered to demystify the legal consultation process for individuals or entities in need of legal guidance. With just a simple PDF document upload, users are ushered into a realm of instantaneous, automated consultations based on the document's contents.

The spotlight of this tutorial is a use case deeply rooted in the Legal domain. Amidst a burgeoning demand for swift and accessible legal consultations, this application emerges as a beacon of convenience. By harnessing the magic of automation and artificial intelligence, it offers preliminary legal advice derived from the uploaded documents, making legal assistance merely a click away.

Concept and Structure of the Application πŸ› οΈ

The essence of this application is to provide a user-centric platform where obtaining legal consultations is a breeze with just a PDF document upload. The collaborative prowess of Streamlit, Vectara, and Langchain serves as the bedrock of this application, orchestrating a robust and intuitive environment.


Streamlit, an open-source app framework, is the mastermind behind the interactive web interface of our application. It fosters a seamless user journey with widgets for file upload and elegantly unfurls the consultation output to the user, making the interface a joy to navigate.

Vectara 🧠

Vectara is the linchpin that elevates the application's capabilities to a new zenith. As a semantic search software company, Vectara is on a mission to redefine search by leveraging artificial intelligence and neural network technologies for natural language processing. It facilitates a deeper understanding of user queries and provides extraordinarily relevant responses. In the scope of our application, Vectara processes the legal documents uploaded by users, delving into the semantics to extract crucial legal insights that form the basis of the automated consultations provided. The integration of Vectara transforms the application into a powerhouse of semantic search, ensuring users receive precise and relevant legal advice based on their documents.

Langchain πŸ€–

Langchain, the engine of text generation in the application, sifts through the legal insights extracted by Vectara to generate automated legal advice. It is the cornerstone that enables the application to furnish text-based consultations, making legal advice easily accessible.

The architecture of the application is elegantly simple yet potent. The user-friendly interface, sculpted with Streamlit, facilitates effortless uploading of PDF documents. Upon upload, Vectara springs into action, processing the document to extract legal insights. Langchain then takes the baton, generating legal advice that is promptly displayed to the user. This synergy ensures that users not only receive instant legal consultations but also have the option for further discussions with legal experts if needed.

Dive into this tutorial to traverse the developmental voyage of this Legal Consultation Application and explore the plethora of features awaiting your discovery!

Setting the Stage: Setup and Installation Guide πŸ› οΈ

Before we delve into the realms of code and explore the intricacies of our application, it's imperative to set the stage right. This segment is dedicated to guiding you through the process of setting up and installing the necessary components for our application. The emphasis is on ensuring a smooth sail as we venture into the development phase.

Step 1: Create a Virtual Environment

Creating a virtual environment is a good practice to manage dependencies and ensure the application runs consistently across different setups.

python3 -m venv vectara-env

Activate the virtual environment:

  • On Windows:
  • On macOS and Linux:
source vectara-env/bin/activate

Step 2: Install Necessary Packages

Install the necessary packages using pip:

pip install streamlit python-dotenv langchain 

Step 3: Create the .env File

Create a file named .env in the root directory of your project. This file will store your environment variables. Here's how your .env file should look like, replace the placeholders with your actual credentials:

OPENAI_API_KEY=openai_api_key here

Step 4: Setup Instructions

Welcome to Step 4! In this crucial phase, we'll walk through obtaining the necessary keys and credentials to kickstart your application. Adhering closely to each instruction ensures a smooth and error-free setup.

1. Navigate to the Vectara Dashboard and Sign In

To begin, open your web browser, navigate to the Vectara Dashboard, and sign in using your credentials.

Navigate to Vectara Dashboard
Open your web browser, head to the Vectara Dashboard, and sign in.

Once logged in, proceed to the data panel, located at the top right, and click to create a new data store.

2. Provide the Necessary Details
Add Data Page

Enter a name and a brief description for your data store.

3. Add Your Data

In the corpora section, you will be redirected to the 'Add Data' page.

Add Data Page

Here, you have the option to manually add data or utilize the code snippet provided later in this tutorial.

4. Access the Control Tab

Navigate to the 'Control Access' tab.

Control Access Tab

Click on the β€œcreate API key” button.

5. Create Your API Key
API Key Creation

Provide a name for your API key, ensuring you select both the Query Service and Index Service.

6. Secure Your API Key

Copy your API key and add it to your .env file.

7. Obtain Corpus and Customer IDs
Corpus ID

Copy the corpus id and integrate it along with the customer id into your .env file. Additionally, secure the client secret by adding it to the .env file.

Importing Necessary Libraries: Setting the Foundation 🧱

Before diving into the intricacies of building our application, it's crucial to import the necessary libraries that will empower our code. This sets the foundation for crafting the interactive web interface, initializing Vectara, streamlining the NLP pipeline, and integrating all these components for a seamless user experience.

import os
import tempfile
import time

import streamlit as st
from dotenv import load_dotenv
from langchain.callbacks.streaming_stdout import StreamingStdOutCallbackHandler
from langchain.chat_models import ChatOpenAI
from langchain.prompts import PromptTemplate
from langchain.schema import StrOutputParser
from langchain.vectorstores import Vectara


With the libraries imported, we now have the tools at our disposal to embark on the journey of creating an intuitive and powerful application.

1. Streamlit: Crafting the Web Interface 🎨

Streamlit is employed to forge a user-friendly web interface for the application, facilitating the easy creation of interactive widgets like text inputs and file uploaders.

# Sidebar for PDF upload and API keys
with st.sidebar:
    uploaded_file = st.file_uploader("Choose a PDF file", type=["pdf"])
    customer_id = st.text_input("Vectara Customer ID", value=os.getenv("CUSTOMER_ID", ""))
    api_key = st.text_input("Vectara API Key", value=os.getenv("API_KEY", ""))
    corpus_id = st.text_input("Vectara Corpus ID", value=str(os.getenv("CORPUS_ID", "")))
    openai_api_key = st.text_input("OpenAI API Key", value=os.getenv("OPENAI_API_KEY", ""))
    submit_button = st.button("Submit")

# Constants
CUSTOMER_ID = customer_id if customer_id else os.getenv("CUSTOMER_ID")
API_KEY = api_key if api_key else os.getenv("API_KEY")
CORPUS_ID = int(corpus_id) if corpus_id else int(os.getenv("CORPUS_ID", 0))  # Assuming CORPUS_ID should be an integer
OPENAI_API_KEY = openai_api_key if openai_api_key else os.getenv("OPENAI_API_KEY")

In this snippet, a sidebar is crafted for configuration, where users can upload a PDF file and input necessary API keys.

2. Vectara: Initialization and Document Retrieval πŸ“š

Vectara is initialized using a simplified function which conceals the intricacies of API interaction, making it beginner-friendly.

def initialize_vectara():
    vectara = Vectara(
    return vectara

vectara_client = initialize_vectara()

In this segment, a Vectara client is initialized with essential credentials which will be utilized to interact with Vectara's services.

def get_knowledge_content(vectara, query, threshold=0.5):
    found_docs = vectara.similarity_search_with_score(
    knowledge_content = ""
    for number, (score, doc) in enumerate(found_docs):
        knowledge_content += f"Document {number}: {found_docs[number][0].page_content}\n"
    return knowledge_content

Here, the get_knowledge_content function abstracts the process of querying Vectara to retrieve pertinent documents based on the user's input query, simplifying interaction with Vectara's neural search capabilities.

3. Langchain: Streamlining the NLP Pipeline πŸ—£οΈ

Langchain is leveraged to establish an NLP pipeline that processes user input and generates responses, abstracting away the complexities of handling language models.

prompt = PromptTemplate.from_template(
    """You are a professional and friendly Legal Consultant and you are helping a client with a legal issue. The client is asking you for advice on a legal issue. Just explain him in detail the answer and nothing else. This is the issue: {issue} 
    To assist him with his issue, you need to know the following information: {knowledge} 
runnable = prompt | ChatOpenAI(streaming=True, callbacks=[StreamingStdOutCallbackHandler()], openai_api_key=OPENAI_API_KEY) | StrOutputParser()

# Main Streamlit App
st.title("Legal Consultation Chat")

# Initialize chat history
if "messages" not in st.session_state:
    st.session_state.messages = []

# Display chat messages from history on app rerun
for message in st.session_state.messages:
    with st.chat_message(message["role"]):

In these lines, various Langchain components are imported and a pipeline (runnable) is established to process user input and generate responses.

4. Integrating and Running the App πŸ”„

The setup integrates Streamlit, Vectara, and Langchain to create a seamless user experience.

if user_input := st.chat_input("Enter your issue:"):
    st.session_state.messages.append({"role": "user", "content": user_input})
    with st.chat_message("user"):

    knowledge_content = get_knowledge_content(vectara_client, user_input)
    print("__________________ Start of knowledge content __________________")
    response = runnable.invoke({"knowledge": knowledge_content, "issue": user_input})

    response_words = response.split()
    with st.chat_message("assistant"):
        message_placeholder = st.empty()
        full_response = ""
        for word in response_words:
            full_response += word + " "
            message_placeholder.markdown(full_response + "β–Œ")

    st.session_state.messages.append({"role": "assistant", "content": full_response})

# Run when the submit button is pressed
if submit_button and uploaded_file:
    with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as tmpfile:
        tmp_filename =

        st.sidebar.success("PDF file successfully uploaded to Vectara!")
    except Exception as e:
        st.sidebar.error(f"An error occurred: {str(e)}")

In this snippet, user input is captured through Streamlit's st.chat_input, Vectara is queried for relevant knowledge content, and Langchain processes the input to generate a response. Additionally, new documents are added to Vectara to enrich the knowledge base when a submit action occurs.

Embarking on the venture of 'The Creation Of Vectara App In Legal Use Case' has been a remarkable journey. The culmination of this project has resulted in a platform that's not just technically sound but one that stands as a beacon of legal assistance for those in need. The Vectara App is designed to be a haven of legal knowledge, making legal understanding and help accessible to everyone.

1. A Visual Tour Through the Interface:

Our application exudes a user-centric design, ensuring that navigating through the multitude of legal information is a breeze. Each section is meticulously crafted to offer a seamless user experience.

Upload your PDF and enter your keys and submit to proceed!

The core of the Vectara App lies in its robust legal knowledge base. Users can dive into a sea of legal information, understand laws, and find answers to their legal queries with just a few clicks.


The expedition of creating the Vectara App in a legal use case scenario has not only been an avenue of technical exploration but a venture into making legal help more accessible. Through this project, a platform has been constructed where legal understanding is not confined to the experts, but is available to all.

The Vectara App stands as a testament to the power of blending legal expertise with cutting-edge technology. As you navigate through the app, the ease with which you can now understand and seek legal help is apparent. This project is a stride towards demystifying the legal realm and making legal assistance a part of the daily lives of individuals.

As we wrap up, the potential for further enhancement and the impact that the Vectara App can create in the legal domain is colossal. The horizon is vast and inviting, promising a realm where legal assistance is not a hurdle but a companion in everyone's life.

Live Demo and Further Exploration

Experience the application firsthand and delve deeper into its mechanics.For a deeper dive into the code and underlying mechanisms, visit the project on Hugging Face:

Try it on Hugging Face

Discover tutorials with similar technologies

Upcoming AI Hackathons and Events