Passport conversations
How Verifable relates public AT Protocol conversation to a Digital Product Passport, without letting conversation become evidence. See publishing to the network for how those posts get written in the first place.
Recognising a linked post
A post that links a public passport page is recognised by that link and enhanced, wherever Verifable renders it, with a linked-passport panel — a visible pointer back to the passport the post is discussing, not a merge of the two.
Conversation as classified statements
Conversation gathered around a passport is recorded as a set of classified statements, each carrying its epistemic kind — question, observation, opinion, proposal, request, offer, commitment, verified-claim, credential-reference, inference, or system-status — and a provenance label. An observation is not an inspection result: someone noting that a product looks worn is an observation, nothing more, until an actual inspection produces evidence.
The current AT Protocol feature keeps that observation in a separate classified statement log. Connecting it onward to a Veritas-backed inspection workflow is the next integration phase. Only a real inspection, with its own proof, may create inspection evidence.
Summaries and source links
Source-aware Veritas summaries are a design target, not a current public service. When wired, a summary must link to its source posts and name a deleted or unavailable source instead of silently retaining the claim without its provenance.
Statement, evidence, and inference stay separate
The intended Veritas integration keeps three graph layers apart: the statement graph (what was said, by whom, with what provenance), the evidence graph (what was actually verified), and the inference graph (what was derived). Today the AT Protocol feature records the first of those as a separate classified log; graph ingestion remains to be wired.