From the moment a request lands to the complete answer.
Four steps. The first three happen in the background, on the schedule you set. The last one runs in under a second, every time.
Connect every place a photo could live.
OAuth into your cloud drives, point us at your CMS, give us read-only S3 keys. Every connector is read-only by default; we never write to your sources.
Crawl on your schedule, never on your servers.
We pull every image touched since the last crawl. Compute happens on EU infrastructure under our control — your sources only see read traffic.
- Incremental: we only re-process images that changed
- Rate-limited: we never push your CMS over its budget
- Audited: every fetch is logged with source URL and hash
Detect every face. Group identities.
Each face becomes a vector. The same person across photos collapses into one identity cluster. Uncertain matches stay uncertain — you confirm or reject during review, and the cluster sharpens.
When a request lands, the answer is already there.
Paste the requester's reference photo or pick the matching identity. Get every appearance, with source URL, file path, crawl date and confidence — exportable as a regulator-ready PDF.
- Search latency: median 0.4s across a 100k-image archive
- Audit trail: every search is logged with the operator and reason
- Export: one PDF, machine-readable JSON, or both
Three pages worth your time.
What Ansikt can detect, group and audit.
Identity clustering, confidence thresholds, the URL proxy, consent tracking.
Compliance, marketing, takedowns, asset discovery.
Built for legal & compliance teams. Quietly useful for marketing.
Article 15 and 17, mapped step-by-step.
How Ansikt maps to the law — and where the law leaves you on your own.