GetRichChat

Landing page
Spotlight
Conversation page

Inspiration

AI-powered financial advisory chat service aimed at aiding customers in analyzing their financial reports and transactions, inspired by Ramit Sethi's approach in Netflix's "How to Get Rich".

What it does

Helps you quickly analyze and summarize the status of your belongings, saving you a lot of money and time, answering the content in the uploaded files in a conversational Q&A format, with future plans supporting lists and graphical statistics, giving you a more intuitive understanding of your belongings

How we built it

Constructed a robust backend using the OpenAI API and Langchain framework in Python hosted on Azure Cloud Service.
Utilized Pinecone, a high-performance vector search database, to streamline the document retrieval performance and efficiency.
Utilized Azure Functions, Azure Cosmos DB, and Azure Storage Blob for efficient data storage and processing.
Applied a deep understanding of Prompt Engineering and Fine-tuning into the service.
Deployed FastAPI to provide reliable API integrations, enabling smoother data interactions and bolstering the service's overall effectiveness.

Challenges we encountered

Efficiently interacting with the Pinecone API for document retrieval and managing vectors (create/update/delete).
Effectively integrating the backend service with the frontend service using Vercel.

Accomplishments that we're proud of

-We enabled document QA at a topic level, which can encompass numerous documents. This approach facilitated conversations that covered multiple documents under a single topic. For instance, after uploading monthly credit card transaction records, we could discuss a financial plan based on those records.