This is a submission for the Hacktoberfest Weekend Challenge: Build for a Friend I built CalmSpace, a small AI-powered reflection tool for a friend or anyone who needs a moment to before reacting.
Sometimes we feel angry, stressed, hurt, or overwhelmed and want to react immediately. CalmSpace gives a simple space to write what is bothering you. It listens to the situation, helps you , suggests one practical next step, and asks one question to understand what you need.
It also has a “Calm my message” feature that rewrites an emotional message into a calmer version while keeping the original complaint and main point.
I wanted to build something simple, private, and easy to use when emotions are running high.
https://github.com/shraddhakolate30-maker/CalmSpace CalmSpace is built with Python and Streamlit. For the AI part, I used Gemma 3:1B, an open-weight model, running locally through Ollama.
The app has two simple features:
Talk to me — the user writes what is bothering them, and Gemma helps them understand the situation, before reacting, choose one practical next step, and think about what they need.
Calm my message — the user can paste an emotional message, and Gemma rewrites it in a calmer way while keeping the main complaint and request.
I chose local inference with Ollama so the AI runs on my own computer instead of depending on a paid cloud AI API.
Open innovation made it possible for me to build CalmSpace using an open-weight AI model that I could run locally.
Using Gemma 3:1B with Ollama meant I could experiment with AI without depending on a paid cloud API. It also helped me understand how a local AI model can become part of a real application, rather than just being something I use through a website.
For a small student project like CalmSpace, this made AI development more accessible and gave me the freedom to experiment, learn, and build with technology that I can run and understand myself.
Best Use of Gemma — $200
CalmSpace uses Gemma 3:1B locally through Ollama as the core AI model.