SScathaPilot Brief

Longitudinal human preference data, collected with consent.

Scatha builds cohort-level, text-only datasets for AI personalization evaluation. Every record is scoped by consent, reviewed for quality, and exported pseudonymously.

Cohort protocolBuyer-use restricted
100Contributor cohort size
4Recurring sessions
0Unreviewed exports
QAExport review status
{
  "cohort_id": "pilot_001",
  "participant_count": 100,
  "session_count": 4,
  "review_state": "qa_pending",
  "export_policy": "pseudonymous_text_only"
}

A protocol for preference signal, not identity resale.

01

Preference signal

Structured responses that make communication style and decision tradeoffs legible.

02

Persona fidelity

Scenario prompts and follow-ups that test whether generated responses preserve intent.

03

Review trail

QA status, consent scope, and buyer-use restrictions attached before export.

04

Export record

JSONL or CSV-ready records with consent scope and dataset card metadata.

Not for high-impact eligibility.

Early pilots should not target employment, credit, insurance, housing, medical, or other regulated eligibility decisions. Re-identification attempts are out of scope.

Text-only v1Consent record requiredPseudonymous export only

Buyer inquiry

Pilot brief requests open after server-side interest capture is connected.

Pilot brief intake not open

This inquiry form is visible for review only. It does not submit or save buyer data until a server-persisted pilot brief endpoint is connected.

Optional. Do not include regulated or high-impact decisioning use cases.