What we sell against these gaps
5 paid products on redhub.ai against eight domains, which leaves 3 with nothing behind them. Those 3 are named below rather than quietly omitted. It is the more useful arrangement: the domains came out of what makes a claim stop being reproducible, and what happens to be on the shelf had no vote in which of them exist.
None of them inspects a system of yours or acts on your behalf. Four grade a description you supply and end in a verdict you can recompute, not a judgment call. The fifth is a subscription, and it is the backbone for a reader nobody outside asks who already has a standing position to keep current.
RedHub AI publishes this site and sells these products. We say so here rather than presenting them as neutral recommendations.
None of these is an audit and none of them certifies anything. They are structured procedures over material you supply, and buying one changes nothing about whether your evidence would survive being asked for.
The 5
Which one closes which gap
A recurring obligation, not another document, which is the only shape that answers a recurring failure. It is the one product here that is not bought once, and the only reason to recommend it is that nothing else on the shelf recurs.
Tracks each risk you have accepted or treated, with an owner and a live review date, so a treatment that has gone stale reads as stale rather than as closed. That is the difference between knowing a fix happened and knowing it held. Its own verdict ladder is where the governance property's bands come from.
Puts each record class against a defined window and a purge-due date, and suppresses anything under a hold from the schedule no matter how far past its date it sits, because a schedule with no way to suspend it eventually deletes something it should not have.
Grades the record itself on what it carries: tool, model, owner, date, source. A missing required field, or a high-impact entry with no reviewer or source on record, forces its worst verdict however complete the log reads on average. Its bands are where this assessment's floors come from.
This assessment's own dry run at higher resolution. You describe the joins for each decision class and it reports, per question, whether the answer is in the record, needs a witness, or has no record at all. It is a self-report like this one and inspects nothing; what it adds is that it asks per decision rather than per organization. The witness idea on your result is borrowed from it.
The 3 we sell nothing against. Each is still asked, and each still carries weight in the score, because an assessment that only asked about the things we sell would be a different kind of document.
When the thing changes. Nothing here closes this one. Noticing that a model version moved or a vendor switched a feature on is monitoring, and we do not sell a monitoring product. The honest first step is a standing item on whoever already reviews vendor changes, not a purchase.
Whether it survives the person. Nothing here closes this one. Whether a procedure is written down, whether a second person has produced the pack, and whether the evidence would still be reachable if somebody left are all things you do rather than things you buy, and none of them costs money. The nearest product on the shelf grades one deliverable before it is handed over, which is a good habit and is not this.
Whether the record can be trusted. Nothing here closes this one either. Whether a record can be altered undetectably is a property of the system holding it, and no workbook can give an append-only history to a folder of exports. This is a question for whoever administers those systems.
Why the AI Audit & Accountability Bundle is not on this page. It exists, it costs $269, and it is cheaper than its four parts, which come to $346 separately. One of the four is on the list above: the AI Output Audit-Trail & Record-Keeping Kit, $79 on its own. The other three are the AI Agent & Connector Access Auditor, which grades what an agent can reach; the AI Disclosure & Synthetic-Content Labeling Readiness Kit, about what you tell people when content is synthetic; and the AI Vendor & Sub-Processor Data-Flow Register.
None of those three answers a question this assessment asks. For a reader arriving from a verdict here, $269 against $79 is not a saving; it is three products bought to get one. If those other questions are live for you as well, the bundle is good value and the arithmetic above is the whole of it.
How the assessment is scored is published in full, weights and bands included.