I started building PDF Analyzer after seeing similar needs in a hackathon. While semantic chunking wasn’t a stated problem, I realized it's a real challenge. Even today, there's no general AI that can semantically chunk large PDFs reliably. This is especially useful for scientific papers. Initially, I used GPT and regex rules (which work for ~50% of cases), but I hit the core issue: no one-size-fits-all logic exists. So now, I'm developing my own AI that understands document structure and performs intelligent, adaptive chunking.
Built With
- mongodb
- nextjs
- openai
- prisma
- shadcn/ui
- tailwind
- xenova/transforms
Log in or sign up for Devpost to join the conversation.