A single call
You send reference and probe in one POST and get the answer. No multi-step flow to maintain.
Send the ID photo and a short video of the face, and it answers whether it is the same person and whether they are live, with the tolerance you define. It stores neither the photo nor the video: it processes in memory and discards.
Comparing two faces is the easy part. The hard part is knowing there is a person on the other side and not a photo of a person held up to the camera. That is why the video is not a whim: it is what rules out spoofing.
All in one system
You send reference and probe in one POST and get the answer. No multi-step flow to maintain.
The person does not have to turn their head or blink on cue. The analysis runs on the video they already recorded.
If the case does not allow video, it accepts a photo. You lose liveness, but the comparison still runs.
The threshold travels in the request or stays fixed per client. A bank and an attendance app do not need the same strictness.
Beyond yes or no, it returns the similarity score, so you can keep the trace or tune your own criteria.
Each consumer gets its own key and its own default tolerance.
Why this one
No photos, no videos, no face vectors. Everything is processed in memory and discarded on response; only the record that a query happened remains, with its score and date.
If no face is detected in the reference or the probe, it returns an error rather than an ambiguous negative. That difference matters when someone disputes a result.
Your system acts on a boolean. Score and liveness stay available for the trace, not for you to interpret.
Multi-tenant by design: the same service serves different products, each with its own key and threshold.
Who it's for
Biometric data is sensitive data in both Chile and Costa Rica. A service that never stores it cannot leak it, and takes your system out of the problem entirely.
Frequently asked
No. They are processed in memory and discarded on response. Only the query record is kept: which system made it, what score it returned and when. Never images or face vectors.
Because a photo cannot tell a person from a photo of that person. Video allows the liveness check to run. If your case does not allow video, it still accepts a photo, but you lose that guarantee.
You do. The tolerance travels in each request or is configured as a default for your system.
Through the contact form. We define the use case, the volume and the tolerance, then hand over the key with the documentation.
Tell us the use case and the volume. It is contracted through the form and we hand over the key with the documentation.