Product
From upload to cited answer — one pipeline
Askivyo is not a chatbot bolted onto a drive. It is an ingestion, retrieval and generation system with tenant isolation in every layer.
How it works
Four stages, one audit trail
Every document, job and answer is attributable. Nothing is generated without a retrieval step first.
Document ingestion
Uploads are stored privately, parsed with structure preserved (pages, headings, sheets, rows), optionally OCR’d, chunked and embedded. Failed jobs retry from the last checkpoint.
Hybrid search
Every query runs vector similarity and full-text search, then fuses the rankings. Authorisation is applied before retrieval, so a result the caller cannot read never enters the candidate set.
Grounded assistant
The model only sees retrieved passages. If the evidence is thin, Askivyo abstains instead of inventing an answer. Tokens stream over SSE so the UI stays responsive.
Verifiable citations
Each claim is bound to a chunk, a document version and a locator (page, section or spreadsheet range). Opening a citation shows the verbatim quote, not a paraphrase.
Retrieval never crosses a workspace
Workspaces are the retrieval boundary. Restricted documents require an extra grant. Cross-tenant reads return 404, not 403 — so the existence of another organisation’s files is not leaked.
Header X-Organisation-Id selects the tenant. Role comes from membership, never from the client.
Row-level security is a second filter on PostgreSQL.
Put your documents to work
Start a free trial, upload a handful of documents and ask the question you have been asking colleagues all week.
No payment details required for the trial.