ListLingo is a voice-powered shopping list assistant that transforms your spoken input into a structured, categorized list organized by grocery aisle — then lets you share it with friends. Powered by the Qwen3:30b-a3b LLM running on Ollama.

Dictate a messy list like:

"Need 2 liters of milk, a couple of avocados, about a kilo of rice, maybe oat milk, and chocolate for Oana."

And get back:

• Milk x 2l
• Oat Milk x 1
• Avocados x 2
• Rice x 1kg
• Chocolate (for Oana) x 1

All intelligently sorted and ready to check off, edit, or send to your shopping crew.

• 🎙️ Voice-friendly natural language input
• 🧠 LLM-powered organization and sorting by aisle
• 💾 Persistent checklists with SQLite
• 🐳 Docker-ready deployment

• Python 3.11+
• Ollama with qwen3:30b-a3b model
• SQLite (included)

Clone the repo:

git clone https://github.com/smfloris/listlingo.git
cd listlingo

Install dependencies:

pip install -r requirements.txt

Start Ollama with the model:

ollama run qwen3:30b-a3b

python app.py

Access the app at: http://localhost:3030

docker build -t listlingo .
docker run -d \
  -p 3030:3030 \
  -e OLLAMA_URL="http://host.docker.internal:11434/api/generate" \
  -v $(pwd)/db:/app/db \
  --name listlingo \
  listlingo

1. 🎙️ Voice input - with Android/iOs voice typing
2. 🤖 LLM parses, categorizes and sorts items
3. 💾 Data stored in SQLite
4. 🔁 API for reading/updating lists
5. 📤 Optional list sharing with friends

MIT — see LICENSE for details.

Almost the whole thing was vibe-coded using Aider.chat with help from Qwen3 and Ollama. This was meant as an experiment to see how far the local models and tooling have come. Code quality-wise, I'm not particularly impressed - but the whole thing works anyway. My own modifications were very minimal and only in one particalar case I HAD to intervene since the AI was going in circles.

Other than the quality of the code itself - the tooling was amazing. Aider is a great tool, Ollama is invaluable and Qwen3 proved to be good junior dev.