How Joring measures AI use
The three dimensions behind the score, what each one captures, and what the number does and does not mean.
Joring scores how well someone works with AI, not how much they use it. Token counts and message volume say nothing about whether the output was any good.
The three dimensions
Joring analyzes each conversation across 61 behavioral signals, grouped into three dimensions.
| Dimension | What it captures | Example signal |
|---|---|---|
| Craft | Prompt quality | Did the prompt carry the context needed to answer it? |
| Command | Human agency | Did the person iterate, or accept the first response? |
| Judgment | Appropriate reliance | Did they verify a claim the model was likely to get wrong? |
No single signal decides anything. A short prompt is not automatically a bad one, and a long conversation is not automatically a good one. Scores come from patterns across many conversations.
What the score is for
The score is a coaching instrument. It exists to answer "what should this person practice next", and it drives which trails Joring recommends.
Why cross-tool measurement matters
A vendor dashboard sees one vendor. If your team uses ChatGPT for drafting, Claude for analysis, and Copilot in the IDE, each dashboard sees a third of the picture and none of them can tell you whether the work got better.
Joring sits across every supported tool, so the same person is measured the same way regardless of which one they opened.
NextPrivacy modelWhat Joring reads to produce these signals, and what it never stores.