README.md

Welcome to Brief.AI

Brief.AI is an innovative platform tailored to hedge funds and investment banks, revolutionizing insights into earnings calls by harnessing large language models.

🤔 Who is Brief.AI?

Our platform aims to be the voice for any executive or analyst on the buy side or sell-side trying to analyze earning call transcripts through two products:

💬 Javi - Question Answering over Earnings Call Transcript

• The intelligent chatbot can engage in real-time queries regarding specific details from earnings call transcripts. This elevates user experience, ensuring immediate access to critical information without manual data trawling.

📃 Long-Short - KPI Extractor

• This model efficiently extracts crucial performance indicators and financial metrics from a comprehensive collection of earnings call transcripts.

🚀 What can this help with?

💬 Chatbot for Earnings Calls:

Our chatbot amplifies the functionality of large language models, empowering users to engage in interactive conversations with earnings calls. This versatile tool serves multiple purposes, such as:

• 🤖 Comparison across multiple documents

  • The chatbot uses an agent to compare queries that retrieves multiple documents and is able to create a chain of thought reasoning chain to answer queries.

• 🧠 Memory:

  • Memory refers to persisting state between calls of a large language model. You can continue to ask follow-up questions from initial queries without restating the context.

• ⚡ Punctual Information:

  • The chatbot provides quick and precise responses to specific questions, making it ideal for extracting timely information from earnings calls.

• 🚨 Sentiment Analysis:

  • Users can gauge the sentiment and emotional tone of earnings calls, helping them make more informed investment decisions.

📃 Detailed Earnings Calls Analysis:

• 📚 Concise answers

  • Utilizing cutting-edge language models, our system delivers succinct, structured answers extracted from verbose earnings call transcripts, streamlining the distillation of key performance indicators (KPIs) for analysts and executives.

• 🧐 Effective KPI extraction from long transcripts

  • Extracting data from unstructured sources like PDFs has become crucial for businesses, researchers, and individuals. Traditional manual methods are slow and error-prone, necessitating more efficient alternatives.

For more detailed information on these capabilities and concepts, please refer to our comprehensive product documentation.

📖 Documentation

For a complete guide to the documentation, please follow the steps outlined below to navigate through the GitHub organization:

• Javi-The-Earnings-Call-Expert: Chatbot powered by langchain.

• LongShort-Dataset: This is the dataset utilized for fine-tuning.

• LongShort: Fine-tuned models designed for KPI (Key Performance Indicators) extraction from earnings call transcripts.

• Website: Platform's UI/UX.

• Branding: Repository containing branding documentation and assets.

🚀 Your Next Stop

• Github: Github organization page.