Holding rate

The percentage of every claim Agent Mode AI has made that is currently marked Holding (still defensible against the latest evidence) as opposed to Partial (substantively revised) or Not holding (falsified or materially overtaken). Computed live across both the AM-NNN enterprise register and the OPS-NNN small-business register, refreshed on every deploy. Surfaced at /holding/insights/ as the publication's primary aggregate trust signal.

The holding rate is the publication's marketable trust number. As of mid-2026 the rate sits in the high band largely because most claims have not yet hit their first scheduled review cycle. The rate becomes more diagnostic over time — the Q3 2026 reading is the first to reflect a corpus where most claims have been reviewed at least once. CIOs citing claims from this publication should reference the permalink at /holding/AM-NNN/ rather than the article URL, because the permalink resolves the claim with its current verdict even after the article has been revised.

Related frameworks

• GAUGE →

Articles that analyse this term

• Q2 2026 Claim Review Bulletin: did the publication's first-quarter claims still hold?
• Why this publication has a ledger — and the analyst sites it benchmarks against don't
• The unverified citation chain: where enterprise AI decisions actually come from

Primary sources

• Agent Mode AI. Holding-up insights — live aggregate dashboard
• Agent Mode AI. Editorial standards