Photos can be spoofed
Screenshots, printed photos and replay from another phone pass weak checks.
Trusted biometric onboarding infrastructure · Kazakhstan
Verify that a live selfie belongs to the same person as a trusted document or profile photo. Face match, liveness, risk scoring, audit trail and partner API for banks, MFIs, pawnshops and e-signature flows.
1-to-1 verification only · no mass identification · no custom model training for MVP
Verification session
matched
0.84
face similarity
0.82
liveness score
low
risk level
1.7s
processing
Problem
A partner may have an application photo and a user selfie, but without a unified process it is hard to prove match, check liveness, store consent and recover audit evidence for security teams.
Screenshots, printed photos and replay from another phone pass weak checks.
Results live in CRM fields, messengers and manual notes instead of a proper audit trail.
Compliance needs scores, reasons, threshold versions and manual review.
Banks, MFIs and signing products need a simple API — not a new ML project.
Solution
01
The partner creates a verification session via API and receives a capture URL.
02
A trusted photo is uploaded from a document, profile or Verex flow.
03
The service checks quality, liveness, extracts embeddings and compares faces.
04
The partner receives decision, scores, failure reasons, callback and audit evidence.
Compare a live selfie with a trusted reference. No 1-to-many person search.
Blink / smile / head turn for MVP, ready for passive liveness later.
Combines face score, liveness, quality, retries and device/IP metadata.
Borderline cases go to reviewers with evidence instead of silent rejection.
HMAC-signed callbacks with retry schedule for enterprise integrations.
Embeddings are never exposed by public API, encrypted and versioned.
Separate keys, rate limits, IP controls and audit logs per partner.
Facex complements Verex ID data and Signex EDS signing instead of replacing them.
Remote onboarding and client confirmation before sensitive operations.
Applications, contracts and collateral tickets with less fraud and manual review.
Connect contract signing to live-client confirmation.
One client verification standard for service points and operators.
The partner system creates a session, uploads reference and live selfie, runs verification and receives a structured result. The biometric service remains internal and is not exposed to the public internet.
POST /api/v1/verification-sessions
X-Api-Key: partner_key
{
"externalUserId": "customer-123",
"purpose": "remote_onboarding",
"callbackUrl": "https://partner.kz/callbacks/biometric-result"
}
→ decision: matched | manual_review | failed_liveness Biometric templates are treated as sensitive personal data and are not returned by public APIs.
User consent, purpose, retention policy and manual review are part of the product — not post-factum documentation.
Facex does not claim government status or certification. Government database integrations require legal authorization.
Before
– selfies sent manually
– no liveness
– decisions not explainable
– audit assembled later
– branches work differently
After
✓ single verification session
✓ face score + liveness
✓ failure reasons and thresholds
✓ audit trail by design
✓ API + callbacks for every branch
Pilot
for first integrations
One partner, API key, demo capture flow, basic admin and manual review.
Business
for networks and products
Multiple partners/branches, callbacks, SLA, retention policy and CRM/1C integrations.
Enterprise
for regulated teams
Dedicated infrastructure, advanced security, legal package and custom liveness model.
No. Facex is commercial identity-verification infrastructure for company workflows.
Not for MVP. We use proven face embedding/liveness components and focus on workflow, API, audit and security.
No. MVP performs 1-to-1 verification: is this live person the trusted reference person?
Yes. Verex can provide structured data and documents; Facex adds the biometric verification layer.
Templates are not exposed through public APIs; they are encrypted and stored with model, version and retention rules.
We will show the demo flow, API contract, admin review and a secure evidence storage model.