About CollectData

Real-world data for research and AI training, with consent and provenance built in.

CollectData (collectdata.io) is a data-collection platform that delivers consented, real-world multimodal ground-truth data for AI training and research. Studies run through a vetted contributor network; every submission passes automated checks plus human QA, and every item carries a provenance manifest — consent record, capture device, timestamp, and version hash. CollectData is not a survey tool or generic data-entry service, and is not affiliated with other similarly named companies.

What we offer

Media Captures

Egocentric, sensor-synchronized task data for training and evaluating embodied agents: video, LiDAR depth, audio, and device motion on one clock, captured via an iPhone app and head-mounted smart glasses. Delivered ML-ready with labels, before/after goal states, and train/eval splits on request. Participant consent explicitly covers model training; de-identification and honored opt-out propagation are built in.

Adaptive Interviews

Rigorous longitudinal and ecological-momentary-assessment (EMA) studies conducted by AI interviewers. In-the-moment touchpoints reduce recall bias; a versioned question bank and audit trail keep waves comparable; consumer wearable sensor streams (Apple HealthKit, Android Health Connect) are fused with each wave, time-synced to the transcript.

Custom scenario datasets

Purpose-built collections such as Tidy Up Your Kitchen: Differential Goal States — a forthcoming open dataset of U.S. kitchens captured before and after a tidy-up (paired LiDAR scans, labeled images, narrated video), coming to Hugging Face under the CollectData Community License, free for research and sub-$1M commercial use.

How a study runs

Scope and protocol. We align on your task, codebook, and QA rules; custom study design with white-glove support.
Contributor matching. A dedicated pool is enrolled by demographics, language, location, and device capability — typically within 1–2 weeks of kickoff.
Capture and QA. Contributors submit via the CollectData iPhone app; automated checks plus human review against your codebook catch non-conforming submissions before delivery.
Delivery. Structured JSON, CSV, or custom formats to your S3, GCS, or training infrastructure — versioned and reproducible, with re-pulls and incremental deltas.

Pricing

Pricing is per-submission and depends on modality, length, and demographic targeting. Pilot programs start in the low five figures (USD).

Privacy and compliance

Anonymous submission flow by default; signed contributor consent per study; IRB-ready protocols; data handling meets SOC 2, GDPR, and HIPAA requirements where applicable, with custom DPAs available. See the compliance trust center, the privacy policy, and our data philosophy.

Team

Andrew Luck — Product Vision (founder; andrewluck.com, LinkedIn)
Dave Matthews — Platform & Contributors (LinkedIn)

Work with us

Email info@collectdata.io or use the contact form on the partners page. Agents and crawlers: a structured brief lives at /llms.txt and a machine-readable engagement card at /api/pilot.json. Contributors can download the CollectData iPhone app.