Taikutsu Lyrz (Salim Noor)
DocuChat

AI web app · PROJECT BREAKDOWN

Your documents. A conversation.

A conversational workspace for PDFs, text files, websites, and YouTube transcripts. Ask questions and explore the material you bring.

PLATFORM
Web · Python
STATUS
Open source
BUILT BY
Taikutsu Lyrz (Salim Noor)

The idea
behind the build.

DocuChat connects document search with a conversational interface. It brings multiple content sources into one workspace so you can move from reading to asking.

WHO IT'S FOR
Students, researchers, and people working with long source material
THE FOCUS
Import content, retrieve relevant passages, and ask conversational questions
01

Start with the material

PDFs, text files, website content, and YouTube transcripts enter a common document workflow.

02

Retrieve relevant context

Embeddings and a vector index provide a way to search imported material by meaning rather than only matching a keyword.

03

Keep creative tools nearby

An image-generation tool sits alongside document interaction for projects that also need visual material.

How the pieces
fit together.

Python web app · Document ingestion · Semantic retrieval · AI tools

The core parts of the project, and the responsibility of each.

PART 01

Browser workspace

Streamlit

Presents upload choices, processing controls, conversations, and image results from a Python application.

PART 02

Source processing

PyPDF2 · BeautifulSoup · Transcript API

Extracts text from the supported document, webpage, and video-caption sources.

PART 03

Retrieval and generation

LangChain · FAISS · OpenAI API

Connects document processing, semantic indexing, chat responses, and image generation.

The workflow

  1. Choose a source

  2. Process the content

  3. Ask a question

  4. Explore the response

The decisions
in the details.

The features, boundaries, and implementation choices that shape the project.

Different source types

Each import route extracts usable text before the material reaches the conversational workflow.

Semantic search

FAISS stores a local vector index used to retrieve passages for document questions.

Python interface

Streamlit supplies the application UI; the project does not use a separate React frontend.

Provider-backed features

Chat, embeddings, and image generation depend on external API configuration and connectivity.

Frameworks & tools.

What the project uses, and where each technology fits.

PythonStreamlitLangChain
Language and UI
Python · Streamlit

Runs the application and provides its browser interface.

AI pipeline
LangChain · OpenAI API

Connects document processing with chat, embeddings, and image generation.

Search
FAISS CPU · tiktoken

Provides semantic indexing and token-handling tools.

Document extraction
PyPDF2 · BeautifulSoup4

Reads PDF text and webpage content.

Other sources and visuals
YouTube Transcript API · Pillow · Altair

Supports captions, image handling, and visual tools.

Availability
& next steps.

How to explore the project, what it needs to run, and the current release status.

Source

The public repository includes Python dependencies and local setup instructions.

Runtime

The app needs a Python environment and external provider configuration.

Demo status

The README currently marks the hosted demo as unavailable. This portfolio page presents the project and its documented architecture.

GitHub repository

Running the app requires the Python dependencies and provider configuration described in the repository. This page introduces the project.