# Sophia depth review

This guide expands the eight-slide Sophia entry point into a forty-slide
learning path at source revision `1ceae52c7896a59ac74a4429d410f3218f0b91c6`.

The sequence keeps five questions separate:

1. Which operation produced the result?
2. What did its live implementation actually do?
3. Which deeper engines exist but are not wired to that route?
4. Which lifecycle and orchestration behavior remains specification-only?
5. What evidence must the consuming surface preserve?

All forty-nine meaningful units in the two pinned V1 Sophia pages are now
reviewed across the existing `sophia-answer-boundaries.json` assignment and the
generated `v1-sophia-depth.json` assignment. Similar prose in the two source
pages remains separately assigned because each page is an independently
reviewable source unit.

The diagrams describe contracts, responsibilities and target relationships. They
do not assert separate deployed services. The live answer continues to use the
in-process Nisaba corpus and deterministic extractive composer by default. BM25,
fact-check and contradiction implementations remain real but unwired to that
customer path; the full ingestion, publication and invalidation flow remains a
V1 specification with some real lifecycle primitives.
