From the card to the catalog in one command.
- aff push ./SDCARD --project warehouse-pick
- Hashed locally, uploaded directly, resumable
- Tag once: scene, task, device, consent
Upload
Product
What it looks like to run your library on Afferent.
Three jobs — store, catalog, deliver — in one place. Every screen below is the product as designed; early access opens to a small group of operators first.
Know what you have, to the usable hour.
- Filter by scene, task, device, status
- Usable hours, not raw duration
- Bulk select → deliver, share, mark licensed
Catalog
Every recording, accounted for.
- Preview with usable-segment markers
- Files with sha256 per file
- Delivery history per session
Session
Hand off 50 TB in hours, not weeks.
- Push to S3, GCS, Azure or R2 — $0 egress
- Signed manifest.json written last
- Or share a link; buyers pull free
Deliver
One link shows your whole library.
- Verified hours and devices from your catalog
- Blurred samples, consent recorded
- Buyers contact you directly. We never sell your data.
Profile
A bill you can predict.
- $9 / TB-month, metered hourly
- Data transfer out: $0, its own line
- Delivery tool: $1 / TB, optional, a compute fee
Billing
What comes after store, catalog, deliver.
We build these in the order operators ask for them — tell us which one matters in the early-access form.
In the pipeline
Validation / QC reports
Usable-hour scoring with blur, exposure and sync flags on every upload.
Automated captioning and tagging
Scene, task and object tags proposed by models, approved by you.
Labelling tools
Turn raw sessions into labelled, training-ready datasets.