| Hey HN, I made a tool to help dog owners understand what their pets are trying to communicate. The problem: Most people can't read dog body language well. A wagging tail doesn't always mean happy, tucked ears can signal many things. My approach:
1. Rule engine scans for keywords (tail, ears, pacing) → instant interpretation
2. AI layer (Pollinations) adds emotional context and actionable tips
3. Breed-aware filtering since a Chihuahua and Great Dane communicate differently
4. History logging so families stay consistent Technical bits I'm proud of:
- Zero-framework vanilla JS that still feels modern
- Smart fallback: AI fails → rules work → offline mode with cached audio
- CSP bypass using CF Worker for embedding restricted iframes
- localStorage-based i18n without bundling MB of JSON It's been helpful for 620k+ households. All client-side, free, no tracking beyond Clarity analytics. Try it: https://dogtranslator.org Happy to discuss the AI prompting strategy or architecture choices! |