The standard

Read every rule before anyone gets certified.

The standard isn't finished yet. Every rule and the full methodology will be published before launch, so you know exactly what a reviewer checks.

No business holds the seal yet. Certification starts after publication.

What the published standard will cover

What counts as generative AI

A plain definition, plus an allowed list for incidental tools such as spam filters.

What customer-facing means

The work, channels and customer conversations a seal covers.

Each tier and its checks

What a reviewer asks for at every tier, from a reviewed file up to monitoring on company-owned devices.

The tool and device list

Which apps let an admin switch generative AI off and keep a usage log, which don't, and which we haven't confirmed. Every entry cites the maker's own documentation and a check date.

What our checks miss

Monitoring has blind spots on each platform. We publish them with test results on the registry, never in fine print.

Decisions and renewals

Who makes the call, how renewal works and what happens when a business fails a check.

Lines already drawn

Some calls are already made. Spam filtering, fraud detection, plain transcription, captions, noise suppression, filler-word removal, background removal and search ranking stay on the allowed list.

Sentence autocomplete, machine translation, sentence rewrites and a vendor's help chatbot all count as generative AI.

At the top tier, every work device has to be company-owned and managed. An app whose generative AI can't be switched off blocks that tier. So does an app whose maker hasn't confirmed its controls in writing.

Why the methodology is public

Your customers should be able to read what the seal checked. So should a competitor who wants to challenge it. We audit against the same published version anyone can read.

Hear when version 0.1 goes public

Waitlist members get the news first, along with founding-member pricing.

Join the waitlist