What it does: Enter a city + your travel style (budget backpacker, luxury, foodie, etc.) → get a personalized PDF guide in ~15 minutes. Not generic "Top 10" lists—actual itineraries with opening hours, transport tips, and recommendations that match YOUR preferences.
Why Perplexity over GPT/Claude: Perplexity's Deep Research API was the game-changer. It actually cites real sources and pulls fresh data. GPT-4 kept giving me outdated restaurant recommendations from 2021. Perplexity searches, synthesizes, and cites—perfect for travel where accuracy matters.
Tech stack:
- Laravel backend with queue workers for long-running generation
- Perplexity Deep Research API for research + synthesis
- Custom PDF generation (tried DOMPDF, settled on Browsershot + headless Chrome)
- Polar for payments (Stripe rejected me 3 times as "travel content")
Hardest technical challenges:
1. API response quality: Prompt engineering to get consistent structure across different cities/personas
2. PDF layout: Making 25-page guides that actually look good and are readable on mobile
3. Queue management: Handling generation failures gracefully + retry logic
4. Payment processors: Finding one that accepts "AI-generated content" businesses
Questions for HN:
1. Would you pay $13 to skip 3-5 hours of research?
2. What would justify higher pricing?
3. How do you feel about AI-generated travel advice vs. human travel bloggers?
Link: https://voyaige.io
Happy to discuss or answer any questions!