Thrift Taste

• 

  screen shot of the app

• 

  screen shot of the app

Inspiration

As a master's student, I've maintained a monthly budget, but the tedious task of manually logging each transaction in an Excel sheet became a challenge. Moreover, my health suffered as I struggled to track my food choices, leading to a 20-pound weight gain. Recognizing the vital importance of managing both finances and health for a happier life, I conceived an AI app called ThriftTaste. This innovative application will let the user update their transaction, responds to expense-related queries, and serves as a diet coach, offering valuable suggestions on my food style. My long-term vision is to seamlessly integrate ThriftTaste with Alexa, enabling convenient and hands-free updates.

What it does

Users can effortlessly update their transactions, inquire about their expenses, and utilize the AI diet coach to assess and improve their dietary habits.

How we built it

I employed the Langchain framework to develop three distinct agents: one for transaction updates, another for responding to expenditure-related queries, and a third for functioning as a health advisor, leveraging the OpenAI model. The application is containerized using Docker, providing easy accessibility and allowing users to run it on their systems with minimal dependencies.

Challenges we ran into

I used llama2 7B quantized model but that still took some time to give results. I tried ollama, but it is not available in windows yet. So, I finally went with Openai model. Initially, the idea was to use speech recognition to update the transactions as manual update is time-consuming, but I could not make the speech recognition work on docker.

Accomplishments that we're proud of

I am proud that the app is working as intended and I did it alone. The app is capable of making anyone's life easier and happier.

What we learned

I'm not very experienced with Docker, but while working on this project, I learned the basics and was pleasantly surprised by how easy it was to run my application on different systems. I'm definitely considering using Docker to deploy my application. Furthermore, as I began developing ThriftTaste, I realized it has numerous possibilities

What's next for Thrift Taste

Find a way to use speech recognition in docker and make the app, voice-enabled.

Built With

• docker
• langchain
• openai
• python
• streamlit