SpeakEasy

DashBoard
HomePage
Conversation Simulation

Inspiration

Have you ever felt uneasy about an upcoming interaction, trying to rehearse and prepare yourself for every deviation a conversation can take? Inspired to help people better prepare themselves and feel confident in their abilities, our team members created SpeakEasy.

What it does

SpeakEasy prepares users for the difficult, and sometimes awkward conversations we all have to face. From speaking to a job recruiter to interacting with your significant other’s parents, we leveraged cutting-edge technologies and machine learning tools to offer personalized assistance to our users.

How we built it

We built our frontend using Next.js, a javascript framework that allows for customization and exhaustive web design. We connected our frontend to our backend code using Flask, a python library. Furthermore, to store our user data and past conversations with the AI, we used MongoDB which allowed us to customize the chatbot interactions. Finally, for the backend, we used python and the NVIDIA NIM API to build our model and allow it to give the user feedback based on their responses.

Challenges we ran into

We had a difficult time integrating the frontend and the backend using Flask. While some behaviors worked, such as the model itself, we could not fully integrate our use of MongoDB into our login, and therefore, our conversations with the AI were not fully custom. Another challenge was designing the website itself as Next.js was a relatively new framework for all of us; therefore, learning the language took us a bit of time. However, we did our best to design the website in a user-friendly way and still integrate it with the python code.

Accomplishments that we're proud of

We are proud of our use of the NVIDIA NIM as it allowed for great user interaction. It gave specific feedback to the user’s response, and it allowed the user to truly learn and grow instead of being put down. Furthermore, we are proud of our integration of the model and javascript, as using frameworks such as flask were also relatively new to us. We hope in the future, we can continue to run Flask so our full functionality of MongoDB can be implemented into our project.

What we learned

We learned how to work with Javascript, and more specifically, ReactJS, Node.js, and Next.js for frontend design. Furthermore, we learned how to work with flask to integrate the frontend and backend of our project, allowing us to develop more complex full-stack apps in the future. Finally, we learned how to work with MongoDB to store user data and integrate it (to an extent) into our AI conversations.

What's next for SpeakEasy

In the future, we aim to better integrate our MongoDB database with our LLM. Although we could effectively store user profile data, being able to store, update, and access users’ previous conversation history and feed it into our NVIDIA NIM model could provide a conversation dynamic that is more representative of a real interaction.