How Face AI Works
A complete walkthrough of the SDK flow from session creation to result delivery.
Architecture Overview
Face AI is a hosted API. Your mobile app integrates our lightweight SDK. Your backend communicates via REST API. No AI infrastructure required on your end.
Your Mobile App
Integrates the SDK
Your Backend Server
Calls REST API with API key
Face AI Platform
Processes, stores, callbacks
The Full Verification Flow
Backend Creates Session
Your server calls POST /api/sdk/v1/session/start with your API key and the user's external ID. You receive a short-lived session token.
Note: Never expose your API key in the mobile app.
Pass Token to Mobile SDK
Your backend passes the session_token to your mobile app. The SDK uses this token — not your API key — for all subsequent calls.
Note: Token expires in 15 minutes.
SDK Captures Media
The SDK opens a camera interface, guides the user through face capture, and uploads the file to POST /api/sdk/v1/session/{token}/upload.
Note: Accepts JPEG, PNG, WebP, MP4, MOV up to 20 MB.
AI Processes the Check
Our servers run liveness detection, forgery analysis, and face comparison in parallel. Average processing time is under 10 seconds.
Note: Processing happens entirely server-side.
Result via Polling or Webhook
Poll GET /api/sdk/v1/session/{token}/status every 2–3 seconds from the SDK. Simultaneously, our server POSTs the result to your webhook URL.
Note: Webhook is signed with HMAC-SHA256.
Understanding Verification Results
approvedLiveness check passed, no forgery detected.
rejectedLiveness check failed — likely a spoofing attempt.
warningLiveness passed but forgery suspected. Review manually.
failedProcessing error — retry the session.