Banking & finance
Open accounts remotely with liveness-backed KYC that resists photos, replays and deepfakes.
How the sector deploys itBehaviorBioAuth (Onboarding)
BehaviorBioAuth confirms that a real, living person is completing an identity check — not a photo, a screen replay or a deepfake. It watches how someone naturally moves, blinks, looks and speaks during a short guided session, then returns a clear pass-or-fail decision you can trust for regulated onboarding
Product Walkthrough
0:000:00
What it is
A short guided session, then a pass-or-fail decision
Each session draws its challenges in a random order — hold still, turn head left or right, follow a gaze target, blink on cue, and speak a number shown on screen while looking at the camera. Because the sequence is never the same twice, a pre-recorded clip cannot be prepared in advance to pass it.
Beyond those explicit challenges, the engine runs continuous liveness checks in the background throughout the session, so authenticity is confirmed from start to finish rather than at a single moment. Onboarding also captures the customer's government ID — Aadhaar or PAN — and confirms the document against the live, verified face, tying a real identity to the person actually present.
The recorded session is examined by a specialised model that looks for the signatures of synthetic or generated video. The fraud verdict is decided on the server, so it cannot be tampered with on the user's device.
- Containerised and self-hosted, so customer data stays within your environment
On-premise
Containerised and self-hosted, so customer data stays within your environment
- The fraud verdict is decided on the server, so it cannot be tampered with on the device
Server-side
The fraud verdict is decided on the server, so it cannot be tampered with on the device
- Agent-led live Video KYC over secure video, alongside the automated 24/7 journey
RBI V-CIP
Agent-led live Video KYC over secure video, alongside the automated 24/7 journey
- PII and biometric templates encrypted at rest, in a format built for hybrid ML-KEM
AES-256-GCM
PII and biometric templates encrypted at rest, in a format built for hybrid ML-KEM
High-level architecture
How BehaviorBioAuth works, end to end
From a short guided session to a server-authoritative decision.
01
Capture
Short guided session on the mobile app
02
Liveness challenges
Randomized actions plus passive, continuous checks
03
Verification engine
Server-authoritative scoring on your infrastructure
04
Deepfake analysis
FaceOff model returns a synthetic-media verdict
05
Decision
Pass or fail, recorded in the admin console
Product exclusiveness
What sets BehaviorBioAuth apart
Behavioral liveness, not a single face match
A simple biometric only asks whether a face resembles the one on file. BehaviorBioAuth observes how a person actually behaves over a short session — the small, involuntary patterns of movement, blinking, gaze and speech that a printed photo, a screen replay or a deepfake cannot reproduce.
Flagship capability
Randomized live challenge library
Each session draws challenges in a random order — hold still, turn head left/right, follow a gaze target, blink on cue, and speak a spoken number while looking at the camera. Because the sequence is never the same twice, a pre-recorded clip can't be prepared in advance to pass it.
Anti-replay · a different order every session
Server-side deepfake analysis
The recorded session is examined by a specialised model that looks for the signatures of synthetic or generated video. The fraud verdict is decided on the server, so it cannot be tampered with on the user's device.
The verdict is decided off the device
Two onboarding journeys, one verified outcome
Run fully automated self-onboarding that is available 24/7, or an agent-led live Video KYC call over secure video (RBI V-CIP) for assisted and higher-assurance cases — both reach the same trusted verdict, so you choose the right level of friction per customer.
Agentless 24/7 · agent-led RBI V-CIP
Passive & continuous liveness
Beyond the explicit challenges, the engine runs continuous liveness checks in the background throughout the session, so authenticity is confirmed from start to finish rather than at a single moment.
Confirmed start to finish, not at one moment
Government-ID capture & face matching
Onboarding captures the customer's government ID (Aadhaar / PAN) and confirms the document against the live, verified face — tying a real identity to the person actually present.
Aadhaar / PAN matched to the live face
Optional vitals & depth modules
When a deployment calls for extra assurance, add optional checks such as heart rate and blood-oxygen (SpO₂) estimation, and a “move closer” depth check that confirms a three-dimensional face in front of the camera.
Switch-on modules
Market comparison
How it compares to conventional onboarding
Proves a live human is present
- Static face-match KYC
- Partly — Single liveness selfie
- OTP / document-only onboarding
- No — Not addressed
- BehaviorBioAuth
- Yes — Behavioral + passive liveness
Resists printed photo & screen replay
- Static face-match KYC
- Partly — Limited
- OTP / document-only onboarding
- No — No
- BehaviorBioAuth
- Yes — Yes
Resists deepfakes & synthetic video
- Static face-match KYC
- No — No
- OTP / document-only onboarding
- No — No
- BehaviorBioAuth
- Yes — Server-side deepfake model
Randomized, unpredictable checks
- Static face-match KYC
- No — Fixed or none
- OTP / document-only onboarding
- No — None
- BehaviorBioAuth
- Yes — Random order each session
Self-service, available 24/7
- Static face-match KYC
- Yes — Yes
- OTP / document-only onboarding
- Yes — Yes
- BehaviorBioAuth
- Yes — Agentless journey
Assisted live Video KYC option
- Static face-match KYC
- No — No
- OTP / document-only onboarding
- No — No
- BehaviorBioAuth
- Yes — Agentic (RBI V-CIP)
Runs on your own infrastructure
- Static face-match KYC
- Partly — Usually vendor cloud
- OTP / document-only onboarding
- Partly — Usually vendor cloud
- BehaviorBioAuth
- Yes — On-premise / containerised
Encryption at rest + PQC-ready
- Static face-match KYC
- Partly — Varies by vendor
- OTP / document-only onboarding
- Partly — Varies by vendor
- BehaviorBioAuth
- Yes — AES-256-GCM, ML-KEM ready
Optional vitals & depth checks
- Static face-match KYC
- No — No
- OTP / document-only onboarding
- No — No
- BehaviorBioAuth
- Yes — Available as modules
The architecture
Hosted by you — one model excepted
One boundary, and only the recorded session crosses it
- Personal data
- ID images
- Biometric templates
- Video recordings
On your premises
Verification engine
Server-authoritative scoring, in containers you host
Trust boundary
Only the recorded session crosses, protected in transit with authenticated AES-256 encryption. Nothing else leaves your premises.
Off your premises
FaceOff deepfake model
The only component not on your premises
AES-256
Pass or fail, in the console
Use cases by sector
Use cases across sectors
Fintech & lending
Onboard borrowers in minutes with a verdict you can defend to auditors.
Telecom
Verify identity for SIM issuance and number portability without a branch visit.
How the sector deploys itInsurance
Confirm policyholders at onboarding and at high-value claim steps.
How the sector deploys itGovernment & public services
Enrol citizens for benefits and services with assisted or self-service Video KYC.
How the sector deploys itCrypto & exchanges
Meet KYC obligations while keeping fraudulent sign-ups out.
Deployment strategy
Runs on your infrastructure. Protects data end to end
Deploy on your infrastructure
Mobile capture, the verification engine, the admin console and the database ship as containers you host on-premise or in your private cloud.
Choose journeys & policy
Enable agentless self-onboarding, agent-led Video KYC, or both, and configure the challenge library and pass thresholds to match your risk appetite.
Connect ID capture & console
Wire up government-ID capture (Aadhaar / PAN) and run day-to-day operations, compliance and fraud review from a single web console.
Deepfake analysis service
Recorded sessions are sent to FaceOff's deepfake model — the only component not on your premises — over authenticated AES-256 encryption, and the verdict returns to your system.
Go live
Decisions are server-authoritative, and sessions, biometric templates and recordings remain encrypted inside your environment.
Data protection
Four things a regulated deployment has to pin down
Where the platform runs, how data is held at rest, what it is future-proofed against, and the one component that is not on your premises.
- Hosting model
- On-premise or private-cloud, fully containerised and self-hosted. Personal data, ID images, biometric templates and video recordings stay inside your environment.
- Data at rest
- PII and biometric templates are encrypted with AES-256-GCM; ID images and Video KYC recordings are encrypted on disk. Searchable fields use a keyed one-way index so values can be matched without being stored in the clear. Encryption keys are mandatory in production.
- Post-quantum ready
- The encryption layer is built for a hybrid ML-KEM (Kyber) + X25519 scheme inside a versioned format, so data is safeguarded against future quantum attacks as the layer is enabled — with no change to how data is stored or migrated.
- Deepfake service
- A single specialised model hosted by FaceOff issues the synthetic-media verdict. Media sent for analysis is protected in transit with authenticated AES-256 encryption; nothing else leaves your premises.
Frequently Asked Questions
It verifies behavioral liveness, not just a single face image. It observes involuntary patterns of movement, blinking, gaze and speech that photos, replays and deepfakes cannot reproduce, and runs passive, continuous liveness checks throughout the session. A specialised model also examines the session on the server for synthetic-media signatures.
Each session draws challenges in random order from a live library — hold still, turn head left or right, follow a gaze target, blink on cue, speak a number. Because the order is unpredictable and combined with continuous passive checks, a pre-recorded video cannot satisfy the sequence. The verdict is decided on the server, so it cannot be tampered with on the device.
Two journeys with one verified outcome: fully automated self-onboarding available 24/7, or an agent-led live Video KYC call (RBI V-CIP) for assisted cases. It can also capture a government ID (Aadhaar / PAN) and confirm the document against the live, verified face. Optional modules add heart-rate and blood-oxygen (SpO₂) estimation and a depth check that confirms a three-dimensional face in front of the camera.
On your own infrastructure. The mobile capture, verification engine, admin console and database ship as containers hosted on-premise or in your private cloud. Personal data, ID images, biometric templates and video recordings stay inside your environment. Only recorded sessions sent for deepfake analysis leave your premises, protected in transit with authenticated AES-256 encryption; nothing else does.
PII and biometric templates are encrypted with AES-256-GCM, and ID images and Video KYC recordings are encrypted on disk. Searchable fields use a keyed one-way index so matching works without storing plain text, and encryption keys are mandatory in production. The encryption layer is built for a hybrid post-quantum scheme — ML-KEM (Kyber) + X25519 — inside a versioned format, so it can be enabled to guard against future quantum attacks with no change to how data is stored or migrated.
The rest of the line
More in BehaviorID
Where it runs
Industries deploying BehaviorBioAuth
Each sector page covers the threat model, the controls, and the regulators that apply.
Verify a living person, not a resembling face
Book a walkthrough and we will show the guided session, the randomized challenge library, the server-side verdict and the console it is recorded in.