Upload a PDF. Get an AI chatbot that knows your docs.
Lexora's knowledge bases handle PDF extraction, chunking, embedding, and retrieval automatically. Free to create — pay only when the chatbot actually answers a question.
PDF + TXT in, chatbot out
Drop a PDF up to 50 MB. We extract text per page, chunk it, embed it with BGE-M3, and store the vectors. Ready to chat in seconds.
Free to create
Knowledge base creation costs zero. You only pay when the chatbot generates an answer. PAYG starts at $0.10 per million tokens.
Citation-aware responses
Every chatbot answer comes back with source citations — page numbers, file names — so users can verify the AI's claims.
OpenAI-compatible API
Pass a `kb_id` in the chat completions call. Works with any OpenAI SDK. No vendor-specific Assistants API to learn.
How simple it looks
Drop-in OpenAI-compatible API. No special SDK.
from openai import OpenAI
import requests
# 1. Upload PDF to a knowledge base
files = {"file": open("contract.pdf", "rb")}
kb = requests.post(
"https://api.lexora.network/v1/kb/kb_my_docs/files",
headers={"Authorization": "Bearer YOUR_KEY"},
files=files,
).json() # → { "file_id": "...", "status": "indexing" }
# 2. Chat — grounded in the PDF
client = OpenAI(
base_url="https://api.lexora.network/v1",
api_key="YOUR_KEY",
)
resp = client.chat.completions.create(
model="Qwen/Qwen3-8B",
messages=[{"role": "user", "content": "What are the termination clauses?"}],
extra_body={"kb_id": "kb_my_docs"},
)
# resp includes answer + sources [{ "file": "contract.pdf", "page": 17 }]How it works
Upload your PDF
Drag-and-drop or POST it via API. Up to 50 MB per file.
Automatic processing
PyMuPDF extracts text and page numbers. We chunk at 1000 tokens with 150-token overlap.
BGE-M3 embeddings (free)
Embeddings generated on Lexora's distributed network. Stored in pgvector under your account.
Chat with citations
Pass `kb_id` to any chat completion. Top-5 chunks injected as context. Citations returned in the response.
Build your first PDF chatbot today.
$1 free signup credit covers your first few thousand chat messages. Knowledge bases stay free forever — you only pay for inference.
Related