Salman Ayman Youssef · Original work
LLM-Powered Website Classification — Actual Sample Output
Developed Python/API workflows and produced domain-classification sample runs using Claude Sonnet. Explored batching, prompt revisions, and webpage-content limits to balance cost, time, and output quality.
Completed in October 2025. The brief targeted up to one million domains. This page shows an excerpt from a development-run output file, with original labels and wording preserved.
| Domain | Classification | Purpose | Confidence |
|---|---|---|---|
| truelearn.com | Corporate | medical exam preparation platform | 95 |
| personalshop.com | Corporate | e-commerce platform | 95 |
| firebase.studio | Corporate | Firebase development tools and services | 75 |
| nic.photography | Indie | personal photography portfolio | 85 |
| myfutbol24.com | Corporate | sports streaming or betting platform | 75 |
| wondercms.com | Indie | open-source flat-file CMS project | 85 |
| scnow.com | Corporate | local news website | 95 |
| grubiks.com | Indie | Rubik's cube solver and puzzle tools | 85 |
These are model-produced labels, not independently verified facts about the websites. Confidence values are model-supplied fields, not measured accuracy.