Tools · General business
Score Explanation
Turn a psychometric score into a plain-language explanation tailored to who's reading it.
How it works
Grounded NLG OVER a computed decomposition (the NLG layer sits above the measurement methods, which stay pure compute). Give it a score decomposition — overall, percentile, reliability, SEM, subscore dimensions, flags — and an audience (candidate / hiring-manager / executive / coach / researcher), and it renders an explanation CONSTRAINED to those facts: no invented drivers, mandatory hedging when reliability is low or a flag is present, and an auditable echo of exactly which dimensions/flags it cited. Same result, different read per audience.
You bring
{ decomposition{ dimensions[], overall?, percentile?, reliability?, sem?, flags?, reference_cohort? }, audience, detail_level?, format?, instrument_context? }
You get
{ audience, explanation, key_points[], what_if[], caveats[], grounded_in (dimensions/flags cited), provenance }
Use it for
- →Give a candidate a supportive, jargon-free read of their assessment result
- →Hand a hiring manager a balanced, evidence-tied summary of what to probe in interview (no hire/no-hire verdict)
- →Explain a low-reliability score honestly — the service is forced to hedge and name the caveat, not present noise as precision
Run it on your data
Call it on your own inputs — over the API, or hand it to your AI agent via MCP. Discovery is open; running it is metered.