AI Product Description Generator (Local LLM)

This is a simple AI-powered product description generator built with FastAPI and connected to a local LLM via Ollama. It allows users to generate marketing-style product descriptions in different tones using models like Mistral, entirely offline and free.

---

Features

• Input product details + select tone (e.g., friendly, technical, sales)
• Generates text using a local AI model
• Works with Ollama — no tokens, no limits
• Built with FastAPI + HTML/CSS
• Includes unit tests

---

Tech Stack

• Python 3.10+
• FastAPI + Uvicorn
• Ollama (with mistral model)
• HTML + CSS
• Jinja2, python-dotenv
• Pytest

---

Setup & Run

Run locally (Poetry)

1. Clone

git clone https://github.com/kat-git-hub/ProductDescriber.git
cd ProductDescriber

2. Install dependencies with Poetry

poetry install

3. Ollama: start & pull a model

ollama serve                    # if not running as a service
ollama pull mistral             # or: ollama pull llama3.2:3b
ollama list                     # verify the model is available

4. Start the FastAPI app

poetry run uvicorn productdescriber.main:app --reload

or, if you use Makefile:

make runserver

Then open in browser: http://127.0.0.1:8000

5. Run tests

poetry run pytest -vv

or:

make test

Run with Docker

1. Build the image:

docker build -t productdescriber:latest .

or:

make docker-build

2. Run the container (macOS/Windows — Ollama on the host):

docker run --rm -p 8000:8000 \
  -e OLLAMA_URL=http://host.docker.internal:11434 \
  -e MODEL_NAME=mistral \
  productdescriber:latest

or:

make docker-run

3. Open: http://localhost:8000

4. Stop the container:

docker stop productdescriber

or:

make docker-stop