ConQuest

An image of the ConQuest webpage

Inspiration

We were inspired by the idea creating a tool to help people reduce their social anxiety by leveraging Generative AI. Our team member Ekin used a mental health service last December. It was not a good experience, while we respect the operator's efforts, the messages were subpar, unhelpful, not tracking the context given by Ekin, and sometimes questionable, like asking about self-harm plans, despite not having any. Ekin had a coincidental visit by our member Nia and our friend Oliver, which was a much comfortable experience compared to the mental health service. The service lacked the human interaction, namely, context and voice. We bring the speech-to-speech service with ConQuest.

What it does

Using ConQuest, users can voice their questions and concerns to an AI assistant. The AI responds back to the user. This method of communication puts the user at ease and provides the users with a starting point on their journey to improvement.

How we built it

We built this project using VSCode, with a mixture of React, CSS, JavaScript, Python, and Flask. Some of us focused on the AI aspect of the project, while some of us focused more on Frontend development. We made it so that we can talk into the website and it will convert our audio into text, use that text for the AI prompt, take the AI response and communicate the response to the user in voice format.

Challenges we ran into

While first developing the AI using Google Colab, we realised that this format does not work directly for a website. We then had to take our Colab code, put it in a plain Python file and adjust the code accordingly.
We were struggling to send the user voice input from the webpage to the python file.

Accomplishments that we're proud of

Used Flask to solve the problem of sending data from the React Frontend to the Python Backend and vice versa to allow a conversation between the user and the AI.

What we learned

We learned how to use the Gemini API and how to prompt it and receive a desired response.
We learned how to use Flask to send and receive data from a React website.
We learned how to use Python libraries to convert speech to text and vice versa.

What's next for ConQuest

We want to implement virtual chatrooms for user with similar levels. This further encourages human connections and progress towards reducing social anxiety.