QuickBrief

AI-Powered PDF Summarization & Q&A Tool

QuickBrief is an intelligent agent that ingests PDFs, extracts key information, generates summaries, and answers questions about the document content. Built with LangChain and NVIDIA AI models, it provides fast, accurate insights from your documents.

---

Features

• Document Ingestion: Loads PDFs and text files from a directory.
• Smart Summarization: Generates concise yet detailed summaries.
• Q&A Capabilities: Answers questions based on document content.
• Vector Search: Uses FAISS for efficient document retrieval.
• Auto-Summarization: Optionally summarizes documents on first interaction.

---

Tech Stack & Models

Core Frameworks

• LangChain (Document processing, retrieval, agent orchestration)
• FAISS (Vector similarity search)
• Pydantic (Configuration & data validation)

Embedding Model

• nvidia/nv-embedqa-e5-v5 (NVIDIA Embeddings for semantic search)

LLM for Summarization & Q&A

• meta/llama-4-maverick-17b-128e-instruct (via NVIDIA AI Endpoints)

---

⚙️ Setup & Installation

Prerequisites

• Python 3.10+
• NVIDIA API Key (for ChatNVIDIA)
• Required packages (langchain, faiss-cpu, pypdf, etc.)

Installation

1. Clone the repository:

   git clone https://github.com/your-repo/QuickBrief.git
   cd QuickBrief

2. Install dependencies:

   pip install -r requirements.txt

3. Set up environment variables (.env):

   NVIDIA_API_KEY=your_api_key_here
   

4. Run the tool:

   aiq serve --config_file note_summarization/configs/config.yml --host 0.0.0.0 --port 8000

---

Configuration

The tool is configured via NoteSummarizationFunctionConfig:

Parameter	Description	Default
ingest_glob	File pattern for PDF/TXT ingestion (e.g., ./docs/*.pdf)	Required
llm_name	LLM reference (meta/llama-4-maverick-17b-128e-instruct)	Required
chunk_size	Text split size for processing	1024
auto_summarize	Auto-summarize on first message	True
embedder_name	Embedding model (nvidia/nv-embedqa-e5-v5)	"nvidia/nv-embedqa-e5-v5"

---

Usage

1. Place documents in the specified directory (e.g., ./docs/).
2. Start the agent—it will auto-summarize documents if enabled.
3. Ask questions like:
   • "Summarize the key points of this document."
   • "What does section 3 discuss?"
   • "List the main recommendations."

---

Troubleshooting

• Error loading files? Check file paths and permissions.
• No documents found? Verify ingest_glob points to the right directory.
• LLM not responding? Ensure the NVIDIA API key is valid.

---

License

MIT License

---

QuickBrief – Turn documents into insights in seconds! 🚀
Built with LangChain & NVIDIA AI.