Imaginal AI

Trust and method

Inspect the evidence system behind the answer

Examine corpus isolation, retrieval, citations, source rights, editions, translations, and provenance before treating an AI mentor as trustworthy.

Topic guide

Use this collection to structure the work

Trustworthy AI mentoring is an information-governance problem before it is a writing problem. The answer should reveal which corpus was searched, which passages were returned, how the source can be located, and when the evidence is too weak to support a claim. When named thinkers are involved, the exact edition, translation, rights status, and permitted territory also determine whether the corpus can be used responsibly.

This collection explains the machinery and governance that sit beneath the conversation. It is designed for readers assessing an AI advisory product, teams planning retrieval architecture, and source owners considering a partnership. The guides distinguish documented evidence from modern interpretation, show why mentor corpora must remain separate, and describe a release process that can keep unresolved works out of production rather than hiding uncertainty behind a polished persona.

Questions this cluster helps answer

Begin with the decision, not the label.

01

How does mentor-specific retrieval prevent corpus blending?

Use the guides below to turn this question into explicit evidence, alternatives, boundaries, and a reviewable next action.

02

What should a useful citation and provenance record expose?

Use the guides below to turn this question into explicit evidence, alternatives, boundaries, and a reviewable next action.

03

Why can an old author still have a rights-blocked modern edition?

Use the guides below to turn this question into explicit evidence, alternatives, boundaries, and a reviewable next action.

Editorial collection

2 substantive guides on trust and method

How to use the library

Read for a decision you can name.

Choose the guide closest to the decision in front of you, complete its diagnostic steps, and record the assumption or action that changes. Follow internal links only when they answer a live question; the collection is designed as a connected method, not a sequence that must be consumed from beginning to end.

When the stakes are regulated, safety-critical, or dependent on facts the system cannot verify, take the resulting brief to a qualified human professional. Imaginal AI can organise questions, evidence, perspectives, and follow-up, but responsibility for the decision remains with the user and the people who hold the relevant authority.