SmartStride

Project Story: E.L.I.T.H Inspiration

The idea behind creating E.L.I.T.H was quite simple for some goal was initially to give people helpful and kind advice that are used for them. It was particularly surprising that there was lack of popular approach guides regarding setting health, finance, and study questions in day to day life. Since three major dimensions of human existence – the mental, the physical, and the financial – define quality of life, it was our mission to create a service that would be accessible by those who could use a reliable and familiar friend every day.

Building the project: What We Learned

In this work, we explored numerous technical disciplines including cloud AI integration and real-time data processing. Key takeaways included:

Working with Google Cloud AI: Next, we discovered how to train and incorporate Google’s generative AI applications, specifically Google Gemini, to reply between:
JSON Data Management: Having a backend in JSON file format let us to store user’s chat history and financial information and keep data storage simple and efficient.
Building with Streamlit: Streamlit became the capable front-end solution for the project as it ensures the scalability and interactivity of the web app.
Teamwork and Task Management: As the tasks were divided among the team members and all the work was done as a collaboration in VS Code, the team members improve their communication, coordinating, and problem-solving abilities.

Building the Project

We designed E.L.I.T.H for users to be able to seek help from the system in real time through a chat based design. Here’s an overview of how we built the project:

- Google Cloud Configuration: Setted up of Generative AI and API keys, to access them Google provided service accounts and credentials.
- JSON Database: To app propriety for conversation histories of an individual or between persons, utilized JSON in order to store chat and financial histories in such manner that actual sequences of the conversation may be useful.
- Streamlit Frontend: Designed a chatbot interface through Streamlit with responsive features like a real time response, message bubbles, and session choices.
- Integration of Google Gemini Model: Employed Gemini to provide users with helpful and informal responses according to the users’ inputs and provide them with valuable insights on learning, income, time, and health.

Challenges We Faced:

No project is complete without its fair share of challenges, and E.L.I.T.H was no exception:

- Integrating AI Responses: The task of achieving reasonable response quality while maintaining reasonable response speed is to become familiar with the model parameters used and to set the error handling.
- Managing JSON Storage: JSON for storing and managing data was not without its problems; how for instance to update records in real-time without creating a merge conflict.
- Real-Time Communication: As for connections it was important to have no delays between components in Streamlit to provide a maximum consistent user experience.
- UI/UX Tweaks: Improving user experience iteratively was difficult while choosing colors, the styling of the chat, and the interactions that were both fun and easy to use.