Drop a document here
PDF · DOCX · XLSX · CSV · TXT · MD · JSON · HTML
1 · Ingestion pipeline
Every stage below is the real backend status — nothing is faked.
Upload
File stored securely in the document vault
Validation
Format sniffed and integrity checked
Extraction
Page-accurate text pulled from the file
Chunking
Structure-aware chunks with page + section
Embedding
Each chunk turned into a search representation
Indexing
Chunks committed to the search index
Metadata
Title, version and page counts recorded
2 · Ask the document
Upload and index a document first to unlock retrieval.
Hybrid retrieval
Keyword + semantic candidates gathered
Reranking
Candidates re-scored for true relevance
Evidence check
Chunks must actually support the question
Grounded answer
Answer written only from validated evidence