Appearance
What this is
The core terms. Mental model: measure a team across CAMS, find the one binding constraint, and apply resource only there — with confidence-gated, interpretable outputs.
CAMS
Capability · Alignment · Motivation · Support. Capable people don't produce performance unless alignment, motivation, and support are present. CAMS is the diagnostic frame Performix scores a team against.
Binding constraint
The single condition that caps performance right now for this team. The diagnostic identifies it rather than listing every possible improvement — so intervention is targeted, not scattershot.
Psychometric-first (AI-as-consumer)
The diagnostic engine is measurement (IRT, MaxDiff, Monte Carlo/EVPI–EVSI, Wilson intervals), precomputed from real psychometrics. AI consumes the substrate around it (schema mapping, evidence extraction, item-candidate generation, ETL) — it never generates the diagnostic. This is the settled framing (the older "AI-first" line inverted it).
The three control levers
- Select-in — who enters the system.
- Develop-in-place — binding-constraint diagnosis + intervention.
- Manage exit-risk — who leaves and why.
Consumer-of-services
Performix composes capabilities owned elsewhere: the CAMS diagnostic runs against the toolbox reincarnation spoke; the Executive Readout consumes Principia measurement methods (competency-rollup + perception-gap); job schema comes from the toolbox job_family_agent. Algorithms never live here — only vendored contracts.
The interaction surface
A small set of executive-facing primitives over precomputed measurement: Executive Readout (framed competency-rollup + perception-gap), Interpretability Rules ("why this nudge/flag"), Conformal-Interval Confidence Gating (decide vs. abstain on prediction uncertainty), and the Coaching Door (simulated-practice scoring + behavioral-signal nudges).
Protected feedback
A CAMS diagnostic isn't a survey screen — it's items selected by reincarnation that surface inline at strategic moments, with feedback kept protected.