Every scored assessment updates who is verified on which work instruction and process, with the evidence one click away. No spreadsheet to maintain.
The AI assessment scores the attempt out of 10 against your approved work instruction. You set the passing score; the default is 8.
A high score alone is not enough. If a step was not observed, the skill is not verified, however good the rest was.
Turn on reviewer sign-off and passing results wait for an admin or reviewer to approve them. Nobody approves their own work.
Set how long a verified skill stays valid for the whole organization, then override it for any work instruction, including never.
When a verified operator fails a later attempt, the skill is flagged and the earlier proof is kept. The case goes to your review queue with the reason, and a person decides.
Publishing a new version never wipes anyone’s status. skillia.AI compares the versions step by step and lists the verified operators whose proof covered a changed step. You decide, person by person, who is reassessed.
Material changes and clarifications are told apart, so a wording fix does not trigger a retraining wave.
Reassess, keep, or mark the change as a clarification only.
Existing proof stays valid until you approve the proposal.
Group work instructions into processes; a process is verified when every one of its work instructions is.
Start with one procedure and a few operators; the matrix builds as assessments come in.