Connect a Library. Stand up a Campaign.
The shared first hour. Every customer story — club, school, news desk, agency — passes through these steps before its own workflow begins. Not a persona; just the product.
Setup is in two stages. Library is one-time and persistent — your sources connected to Ansikt, your photos readable. Campaign is per-engagement — a named, time-bound scope that wraps photos from your Library with a legal basis, Purposes, Channels, and an end date. One Library, many Campaigns over time.
The DPA you signed at signup is already in force. It is not part of this walkthrough.
Today the demo shows the Library screens. The Campaign management screens are Planned — the conceptual model below is canonical; the screenshots will catch up.
One-time. Persistent.
Connect your Library
Authorise read-only connectors against your sources — DAM, SharePoint, Google Drive, S3, your CMS. Ansikt becomes your processor against those sources; you stay the controller. The DPA you signed at signup covers this; this step just operationalises it.
Off by default, from day one. Any face Ansikt cannot positively match to a consent record is blurred when the proxy serves the image. Non-responders are Blocked until they decide.
Ansikt does not yet compute a persistent biometric template for any subject — that is the numeric face reference Ansikt stores once a subject consents, used to match the same person across photos. Templates wait until each subject themselves consents inside a Campaign.
Nothing publishes yet. Publication waits for a Campaign — that is Stage 2. The audit history line for the connector authorisation is dated and attributed to the operator who set it up. If anyone three years from now asks "when did you connect that source?" — the answer is on screen.
Per-engagement. Time-bound.
Create your first Campaign
A Campaign is a named scope around a specific engagement — a season's match-day photos, a campaign shoot, a clean-up of an old archive. Name it. Pick the legal basis — consent, legitimate interest, or public task. Set an end date; every consent inside this Campaign expires on that date regardless of when each subject signed. Pick the image scope: the folders and files in your Library that this Campaign covers. In preview
One Library, many Campaigns. A subject who appears in three Campaigns is the same identity in your Library, with three sets of decisions. Subjects see their full set of Campaigns when they open their portal.
Define Purposes and Channels for this Campaign
Purposes are what subjects consent to — specific uses with specific scope: "match-day Facebook posts", "yearbook", "annual report", "the quarterly newsletter". Not "marketing." Not "communications." Subjects decide per Purpose.
Channels are the destinations under each Purpose — the actual Facebook page, the printed yearbook book, the website CMS. Subjects see the Channel list at sign time as transparency. The operator picks the Channel at the time of each publish. Channels are documentation, not consent dimensions.
Each Channel splits by takedown route. Most are manual: Ansikt records what was published where, but the takedown on a withdrawal happens in the destination's own UI by someone on your side. A smaller set are eligible for the URL Proxy In preview: you serve your existing image URLs through Ansikt, and a withdrawal blurs only the withdrawing face on the mediated URL. The proxy is opt-in per Channel; until you opt in, every Channel is manual.
First publication — Image × Purpose × Channel
Pick an image from your Library that falls inside this Campaign's image scope. Ansikt detects each face on the page. The operator taps a face, types a name and an email, ticks the Purposes this Campaign is asking the subject to consent to, and sends the request.
Until that subject says yes, two things are true. The face exists in Ansikt as a per-photo detection only — no persistent embedding has been stored, so the same face on another photo is not auto-matched. And the photo cannot be published under that subject's contribution.
When the subject opens the magic link in their portal and grants consent, the biometric template is computed and persisted; from that moment, the same face is recognised across your photos. The first Publication record — Image × Purpose × Channel, one row in the audit log binding one image, one consented Purpose and one Channel where it was published, dated and attributed to the operator — lands in the audit log. Every subsequent publication threads through the same record.
That is setup. Library connected. First Campaign running. From here, every customer story picks up its own workflow — the Saturday match-day post, the school class photo, the end-of-season withdrawal, the Article 15 export.
That's the shared starting point. Pick the story closest to your shape.
Five personas — two customer-side, three subject-side — each picking up after the Library and a first Campaign are running. Read whichever maps to your own situation, or skip back to the tour index.