Features & Architecture
How LiteChatbot turns your documents into answers: chunk, embed, retrieve, reply.
Document & URL RAG engine
Upload PDFs, DOCX or paste website URLs. The engine parses documents into chunks, generates vector embeddings and stores them in PostgreSQL pgvector. At chat time it retrieves only the most relevant chunks, so answers stay grounded in your content instead of hallucinating.
Smart model routing
| Model | Best for | Speed |
|---|---|---|
| DeepSeek-V3 | Everyday support, ultra-low cost | Fastest |
| DeepSeek-R1 | Multi-step reasoning, troubleshooting | Slower |
| GPT-4o-mini | Dependable general fallback | Fast |
Lead capture & routing
When the bot can't answer, visitors leave name + contact with a consent tick. The free WP plugin emails you instantly; the SaaS dashboard adds WhatsApp owner notifications.
Security & isolation
Embeddings are compartmentalized per tenant (tenant_id). Anti-abuse fingerprinting hashes site_url + admin email server-side, so free quotas can't be reset by reinstalling.