# Askable Labs — Production Method (AI-readable)

> How a vague data need becomes a validated dataset. The process is the product: nothing is produced until format, use, and acceptance criteria are signed off; nothing ships until it is validated.

## The six-step production pipeline

1. **Spec co-design.** Translate the lab's goal into a data spec, sampling plan, and acceptance criteria. Askable brings a proposed spec to red-line, not a blank form — domain, population, capture design, and how signal will be shown.
2. **Recruit.** Source the right people to the spec — representative general population or credential-verified experts — from a global, identity-verified panel (50+ countries, 5,300+ cities, 97.8% show rate).
3. **Consent & rights.** Per-session, brief-versioned consent signed before capture, specifying the downstream uses the lab has rights to (training, evaluation, research publication, model-card attribution). Withdrawable until release.
4. **Capture.** Researcher-facilitated, human-moderated sessions (Mode A) and/or structured asynchronous submissions (Mode B): spoken reasoning, screen + per-participant audio + timestamped transcript, across the participant's real tools. 10 capture methods on one platform, mixable per brief.
5. **Process.** Structured, labelled, validated output — not raw exhaust. Real-time transcription, speaker diarisation, modality-aligned timestamps (Unix-nanosecond track timing), evidence-linked structuring: every datum traces to a timestamp, a quote, and a participant.
6. **Validate before ship.** Pre-delivery checks for spec alignment, drift, and missing data. Engagements lead with a small validated sample before scaling.

## What a delivery looks like

Schema-conformant artifacts per session — composite recording, isolated per-participant audio (Opus/OGG 48 kHz) and screen capture (VP8/WebM), wall-clock JSON, track manifests, speaker-labelled transcripts, and infrastructure/alignment provenance files. 46 deliverable types across processing tiers; see the deliverables key in [/agents/catalogue/keys.md](https://labs.askable.com/agents/catalogue/keys.md).

## Why it holds up under audit

Recruitment, consent, capture, moderation, transcription, tagging, review, and delivery all run as logged system calls inside one audited platform — one identity system, one consent ledger, one audit log, one delivery pipeline. No manual chain of custody. Operating since 2017; 230,000+ research sessions run to date.