JobFrame · Coordinates
JobFrame is Pantone for jobs: a trusted canonical reference where every role has a number and computable coordinates.
So similarity, leveling, and pricing become measurable instead of arguable. Before Pantone, “match this red” was an argument; a number plus a calibrated reference ended it. JobFrame does the same for work.
See it — a role and its nearest neighbors
Real data. Distance runs 0 (identical) → 1 (unrelated); each neighbor carries the 3-state band against tolerance. This is structural distance — same architecture, the level ladder dominating.
One role, several coordinate spaces
Like a colour in LAB vs CMYK vs RGB — same swatch, different systems for different questions. We keep them separate on purpose: two roles can be close in pay but far in content, or the same work at different levels.
Structural
liveWhere does this role sit in the architecture?
level ladder + family / function / focus
Content
liveWhat work does it actually involve?
skills, knowledge & activities from public O*NET
Pay
liveWhere does it sit on the pay surface?
modeled pay surface from the combined survey-blend fit
Semantic
liveWhat does its description mean?
frozen embedding vectors of the profile prose
Why you can trust it
How matching works →It says “I don’t know.”
Every comparison returns confident, needs-review, or no-confident-match against a tolerance — below the bar JobFrame refuses to guess rather than fabricate a match.
It’s frozen in editions.
Coordinates belong to a dated, versioned edition. A number computed today still means the same thing next quarter; improving the math releases a new edition.
It’s built from real data.
Structural coordinates come from the canon; content coordinates come from public U.S. O*NET data. Nothing is invented.
Explore the canon at /jobframe/canon or query coordinates directly: GET /api/spokes/job-family-agent/coordinates?profileKey=…&neighbors=10