Evidence intake sessions

An evidence intake session is how documents become structured proof. You bring in artifacts, the session processes them through a fixed sequence, and the result is traceable evidence mapped to your capabilities, without Pytheus retaining the original files.

The processing sequence

A session moves through defined stages, each with a real backend state rather than a simulated progress bar. In order, it receives the bytes, runs security checks, identifies document types, extracts text, identifies facts, maps them to capabilities and questions, validates coverage, persists the structured evidence, and then deletes the original document and verifies the deletion before finalizing outputs and preparing the review.

Originals are erased

This is a deliberate privacy design. Once facts are extracted and persisted, the original document is deleted and the deletion is verified. Pytheus keeps the structured evidence and traceable references, not your document content. A document is never reported deleted before its erasure is confirmed.

Approval is the only handoff

Processing a session does not change your assessment. Only when you approve a session are its answers materialized and handed off to Pytheus Intelligence for scoring. Absence of evidence is treated as Unknown, never as a No, and applicability is tracked separately from the answer itself.

Nothing in intake silently moves your score. Extraction produces a reviewable result; your approval is the deliberate step that lets it count.

Approved evidence shows up as documents in the Evidence library and shifts controls between verification states. Sessions and their status are also visible under Integrations.