SPARROW - Smart Proposal Assistant for RFP Response and Offer Workflow

Project Documentation

Quick Overview

The project aimed to streamline the RFP process by automating the review and creation of client proposals. Utilizing the latest 247-page STIHL Catalogue 2024, we processed product descriptions with Langchain chunking and OpenAI Embeddings, storing them in a Pinecone Vector Database. Synthetically generated RFPs by GPT-4 were formatted into multiple file types, with key details extracted into JSON and similarly embedded. A RAG pipeline facilitated the retrieval of top product matches, leveraging GPT-4 for final selections, which were then composed into client emails using Microsoft's PHI3.

Demo

aaltoai_hackathon_sparrow.mp4

Pipeline Overview

Installation & Setup

Create Virtual Environment

python -m venv hackathon_ai

Activation

• macOS/Linux:

  source hackathon_ai/bin/activate

• Windows:

  .\hackathon_ai\Scripts\activate

Installing Dependencies

Before running the project, you need to install all dependencies. To download the requirements.txt file, use the following command:

pip install -r requirements.txt

Additionally, to successfully run the web application as well as the scripts and notebooks that use the OpenAI API and the Pinecone API, you need to paste your own API key from OpenAI and from Pinecone into the .env file:

OPENAI_API_KEY=<YOUR_OPENAI_KEY>

PINECONE_API_KEY=<YOUR_PINECONE_API_KEY>

Web Application (Streamlit)

To run the web application after you have installed all the dependencies, you can run this command down below from the project root directory (make sure you added the OpenAI key in the .env file!):

streamlit run app/app.py