Praxi

• 

  Homepage

Inspiration

We initially developed the idea for Praxi in a similar way to how many projects start off - we weren't 100% happy with what was out there right now, and thought that we could make something new to fix it. The emergence of AI Chatbots in the last few years also seemed useful for us to leverage as a practice tool, and our research confirmed that the best way to learn a new language is talking to the locals themselves - thus we have Praxi, pronounced 'prax-eye', derived from the word praxis.

What it does

Praxi is based on two core features: live and AI conversation. The AI chatbot is used to converse with the user and serve as practice by providing quizzes, assisting to learn new language components (e.g. a new verb tense), and more. Live conversation is intended to serve as a test for users, where they are able to measure there skills against native speakers.

How we built it

Primarily a django (Python-based) web application with HTML and TailwindCSS pages to serve as the frontend. PostgreSQL in the back to handle the data & messages, as well as web sockets to handle the chat feature. The Chat bot was created using the ChatGPT API.

Challenges we ran into

Dealing with view-based Django applications and attempting to use React / Next.JS with it, as well as all of us being relatively new to web sockets for the chat aspect. Definitely a great learning experience.

Accomplishments that we're proud of

Overall, we are proud of our UI design, use of new technologies (Django, web sockets, ChatGPT API), and general progress from our first Deltahacks last year to now.

What we learned

As usual with a hackathon, we all learned a great deal in a short time. We learned about how a backend selection can limit the frontend, how to change major components of a project on the fly, as well as a healthy amount of Tailwind.

What's next for Praxi

We hope to continue to add features - a progress tracking tool that could measure time spent with the practice bot and conversing with native speakers would be a great addition. An additional idea we had would be allowing those who have already mastered a language and are fluent to provide a score / feedback to those who are still learning - this will only increase the learning efficiency available on our platform.

Built With

• django
• html5
• javascript
• postgresql
• python
• tailwindcss
• websockets