Tools · People analytics
Exemplar Classifier
Classify any free-text response against labeled exemplars — predicted class + calibrated confidence.
How it works
Public-domain implementation of US20180025303A1 (abandoned patent). Two channels blended per the patent's multi-dimensional framing: (1) a semantic embedding channel (key-gated — real provider embeddings with OPENAI_API_KEY, deterministic lexical fallback without, flagged `degraded`, never silently faked) and (2) a psycholinguistic channel of textbook constructs — Weintraub's 14 speech categories (1964) + Mehrabian's psychological-distance/nonimmediacy (1966), all deterministic/key-free. Nearest-prototype clustering against the labeled exemplars; softmax-calibrated confidence; full feature + nearest-exemplar explainability.
You bring
{ response, exemplars? | exemplarSetId?, semanticWeight?, topK? } — inline labeled exemplars OR a built-in set id ('interview-orientation')
You get
{ predictedClass, confidence, margin, classScores[], nearestExemplars[], features{weintraub,mehrabian}, embedding{mode,degraded} }
Use it for
- →Score a constructed interview response against ownership-vs-deflecting exemplars (built-in 'interview-orientation')
- →Bring your own validated, labeled exemplars → segment open-ended survey/assessment responses with provenance
- →Inspect the Weintraub/Mehrabian profile of any text — self-reference, hedging, psychological distance — as standalone features
See it work
example outputA candidate's constructed interview answer scored against the built-in 'interview-orientation' exemplar set (ownership vs deflecting).
Exemplar Classification — interview-orientation
Response scored: "When our launch slipped, I owned it. I rewrote the rollout plan myself, called the two biggest customers, and shipped the fix that week. I'd rather catch a problem early and answer for it."
Verdict
- Predicted class:
ownership - Confidence: 0.86 (softmax-calibrated)
- Margin over runner-up: 0.72
Class scores
| Class | Blended | Semantic | Psycholinguistic | Support |
|---|---|---|---|---|
| ownership | 0.86 | 0.84 | 0.89 | 4 exemplars |
| deflecting | 0.14 | 0.16 | 0.11 | 4 exemplars |
Nearest exemplars (explainability)
own-2(ownership, sim 0.81) — "When the project stalled, I stepped in. I rewrote the test plan… and I shipped the fix the same week."own-1(ownership, sim 0.78) — "I took responsibility for the missed deadline. I called the client myself…"own-4(ownership, sim 0.74) — "I volunteered to lead the migration… I am accountable for that decision."
Psycholinguistic profile (query)
- Word count: 41
- Weintraub (rate /1,000 words): high
iself-reference, highdirect_references, lowqualifiers/retractors— agentic, un-hedged. - Mehrabian non-immediacy: 0.18 (low distance) — active voice, first-person, present-framed; little objectification or passivity.
The psycholinguistic channel agrees with the semantic channel: first-person agency and concrete action, the structural signature of the ownership exemplars rather than the impersonal/passive deflecting ones.
Embedding channel
- Mode:
api· modeltext-embedding-3-small· semanticWeight 0.6 ·degraded: false(real provider embeddings; the lexical fallback would set this true).
Provenance: service exemplar-classifier · patent US20180025303A1 (abandoned → public domain) · method weintraub-1964 + mehrabian-1966 + embedding-exemplar-match · exemplarSet interview-orientation. Built-in set = illustrative demonstration labels, not validated psychometric norms — bring your own labeled exemplars for a clinical-grade signal.
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.
Related questions we answer
- How do I measure employee performance at work?Separate the signal you want — the person's actual contribution — from the noise that dominates raw ratings, chiefly idiosyncratic rater variance, which often explains more of a score than the ratee does. Define the construct before the scale, use behaviorally anchored items, calibrate raters, and report reliability. Measure conditions and outputs, not personalities.
- How mature is our people analytics program?Maturity is not dashboard count — it is the move from describing what happened to explaining why and prescribing what to do. Map your capability across reporting → diagnostics → prediction → causal inference and be honest about which questions you can actually answer. Most 'programs' are reporting with a BI tool.
- What is the difference between people analytics and HR reporting?HR reporting describes what happened — headcount, turnover rate, time-to-fill. People analytics applies behavioral science and causal methods to explain why and prescribe action. It is a different epistemology, not a prettier chart: reporting counts, analytics infers.
- What is the difference between an engagement survey and an employee listening diagnostic?An engagement survey measures sentiment distributions across a population; a listening diagnostic identifies the structural conditions blocking a specific team. They answer different questions — 'how do people feel overall?' versus 'what is binding here, and what is the one action?' — and require different methodology and sampling.
- Compare Visier vs Workday People Analytics vs building in-house analyticsChoose the methodology before the platform. Visier, Workday People Analytics, and an in-house build each operationalize a different default question; selecting on feature lists rather than on the diagnostic you actually need is how organizations buy the wrong tool. Decide what you must measure and how, then pick what serves it.
- What are the alternatives to annual employee engagement surveys?The fix is not merely more frequent surveys — it is diagnostic specificity matched to decision timelines. Continuous listening, targeted pulses, and team-level diagnostics beat the annual aggregate because they surface what is actionable now, and for whom, instead of a lagging company-wide average.