A conversation record needs more than a final answer
The Conversation Index is a reading method for model exchanges that circulate in screenshots, reports, support tickets, papers, and public debates. It asks how the exchange moved, what context was visible, which repair moves mattered, and whether the final answer deserves to stand alone. This page is not a chat archive. It is a way to preserve the shape of an exchange so a later reader can understand why an answer felt convincing, brittle, evasive, useful, or misleading.
Does the exchange move from task to correction, role-play to refusal, or evidence request to unsupported synthesis?
Which pieces were visible to the model, visible only to the interface, or inferred by the reader after the answer?
How does the user steer, reject, narrow, or reframe the answer after the first output?
What can a later reader learn from the sequence that the final answer alone would hide?