This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady
What I Built
Fluently — Your Personal English-Learning Companion
Learning English consistently can be challenging when you need to manage classes, track attendance, and find time for daily practice. Fluently is designed to bring these activities together in one place and make building an English-learning habit easier.
Fluently aims to help learners organise their learning routine, practise English regularly, and get help from an AI teacher. Instead of spending too much time deciding what to practise next, learners can focus on improving their communication skills, one step at a time.
The idea is simple: make consistent English practice easier, so learners can spend less time organising and more time learning.
Demo
🚀 Try Fluently: http://fluently-4lhh.onrender.com/
Code
💻 GitHub Repository: https://github.com/PalakshiVerma/Fluently
How I Built It
Fluently is built around Python and Streamlit, with the goal of integrating an AI teacher powered by an open-weight Gemma model through a hosted inference API.
The project architecture uses the following technologies, where implemented:
- Python + Streamlit: Application logic and user interface.
- Gemma: AI-powered assistance for English-learning questions and follow-up conversations.
- SQLite: Structured data such as classes, attendance, and practice progress.
- PyMuPDF and python-docx: Extracting text from learning materials, if the materials feature is included.
- ChromaDB and embeddings: An optional retrieval layer for answering questions based on uploaded notes.
- Render: Hosting the web application.
The focus is on using AI to make English practice more accessible and easier to maintain, rather than building a generic chatbot without a specific purpose.
Why Does Open Innovation Matter?
Open innovation gives developers the opportunity to experiment, learn, and build specialised solutions using openly available models and tools.
For Fluently, open-weight models such as Gemma offer a way to explore how AI can support language learning and natural conversations. Open-source frameworks also make it possible to build and customise the surrounding application without starting everything from scratch.
I am using a hosted inference approach to make development practical without needing to run a large model on my own hardware. This is an important distinction: using an open-weight model through a hosted API is different from running the model locally.
For me, this project is about learning how to turn AI capabilities into something useful for a real problem—and making the result accessible through a simple web application.
The Lazy Payoff
Fluently aims to reduce the friction between wanting to improve English and actually practising it.
By bringing learning activities together in one place and offering AI assistance, the project is designed to make it easier to maintain a routine without spending extra effort organising every session.
The intended payoff is simple: less friction, more consistent practice, and more time spent improving communication skills.
Prize Categories
- Gemma: If the deployed application uses a Gemma model as described above.
Top comments (0)