Free diagnostic
The Column Test
Ten questions, run on your own AI governance, scored out of 100. You get a card showing where you stand, which dimension fails first, and what closing it involves. Three minutes, no email required, and you can print it and take it into your next meeting.
The interactive version needs JavaScript. The ten questions, to run by hand — score each 0, 1 or 2:
- Literacy. Have the people deciding on AI output been trained to judge it, role by role, with a record?
- The override. Is the person who catches the model being wrong named, trained, and on the record having overridden it?
- Succession. If that person left tomorrow, does a named second already do the same work?
- Ownership. Does each production AI system have one named accountable owner who can stop it?
- The halt decision. Does switching a system off sit with a named role against a threshold set in advance?
- Board cadence. Does AI risk reach the board as a standing item with a number that moves?
- Vendor inventory. Is the list of third-party models and AI vendors current, owned, and clear on what each can see?
- Vendor change. If a vendor changed their model underneath you, would you catch it?
- Funding. Are literacy and governance funded as a standing line rather than a project with an end date?
- Evidence. If a regulator asked on Monday, would the evidence already exist?