A public register of claims the industry is making about enterprise AI.
Every week, vendors, analysts, academics, and tier-1 publications make specific, testable claims about agentic AI. Most are never tested again. This archive logs the significant ones with timestamps, source snapshots, and a scheduled review. Verdicts move as evidence moves. How this works → Claims made by this publication — are tracked separately at /holding/.
Spotted something worth archiving? Submit a claim candidate → Every submission is triaged against the significance bar; passes land in the archive with a Wayback snapshot.
Aggregate view: verdict distribution, source-type breakdown, activity heatmap, status-change timeline →
Want the raw data? Quarterly CSV and JSON bundles arrive at /archive/data/ starting Q2 2026. RSS feeds at new claims, new reviews, and status changes.
Reference format for researchers, journalists, analysts.
The archive is registered as a schema.org/Dataset so it surfaces in Google Dataset Search and AI-search research surfaces. The machine-readable Dataset record is at /archive/dataset.json. Individual claims carry their own per-row Dataset entity; see any claim's <script type="application/ld+json"> block.
Agent Mode AI. (2026). Agent Mode AI Claim Archive: A public register of enterprise-AI claims [Dataset]. https://agentmodeai.com/archive/
@misc{agentmodeai_claim_archive,
author = {Agent Mode AI},
title = {Agent Mode AI Claim Archive: A public register of enterprise-AI claims},
year = {2026},
howpublished = {\url{https://agentmodeai.com/archive/}},
note = {Dataset. Methodology: https://agentmodeai.com/archive/method/}
}Agent Mode AI Claim Archive, AM-NNN, retrieved 2026-06-03. https://agentmodeai.com/claims/AM-NNN/
Prefer a programmatic cite? Every claim page carries a Dataset JSON-LD block at /claims/AM-NNN/ that contains the claim id, source, snapshot, verdict, and review history as a row-level record. Attribution style: whatever your style guide prefers — URL, claim ID, and retrieval date are sufficient.