Learn how a Video KYC Session Assistant AI Agent guides customers, checks liveness and documents, captures geo-tags and audit trails and lifts V-CIP productivity.
A Video KYC Session Assistant AI Agent is an AI system that helps lenders run compliant video KYC sessions. It guides customers, checks liveness and documents, captures geo-tags and audit trails, and helps officials complete more V-CIP calls per hour.
RBI introduced the Video based Customer Identification Process, or V-CIP, in January 2020. It allows banks and NBFCs to complete full KYC remotely through a live video interaction.
V-CIP has become central to digital lending, but the process is demanding. Each call must follow strict steps, and failed or incomplete calls waste official time. The agent makes each session faster, more reliable and easier to audit.
The agent is a co-pilot for the V-CIP official and a guide for the customer. It runs checks, prompts each step and records evidence, while the official conducts the interaction and makes the decision.
V-CIP officials, onboarding operations, compliance, concurrent auditors and fraud teams.
It sits inside the digital onboarding journey and builds on established practice in video KYC for lenders. For loans, it usually runs before decisioning is finalised and agreements are signed.
Video KYC is both a growth enabler and a compliance risk. Productivity and quality need to improve together.
The KYC Master Direction sets specific requirements for the official, the technology, the recording, location checks and audit. Missing any step creates compliance risk.
Every session needs a trained official in real time. Long or failed calls directly limit how many customers can be onboarded.
Poor networks, bad lighting, missing documents and unprepared customers cause drop-offs and repeat calls.
Deepfakes, replay attacks and impersonation target remote onboarding. Deepfakes in fintech are a growing concern.
Accounts opened through V-CIP become operational only after concurrent audit. Slow audits delay disbursement.
Customers who face long waits or repeated calls abandon onboarding and choose another lender.
The agent supports the session before, during and after the call.
The customer gives consent for V-CIP, and the agent offers a suitable time slot or immediate connection based on official availability.
It checks network speed, camera and microphone and reminds the customer to keep their original PAN card or e-PAN and other documents ready.
Aadhaar offline XML or e-KYC data is fetched and validated. RBI expects Aadhaar XML used in V-CIP to be no older than three days, and the agent checks this.
During the call, the agent guides the customer on positioning and lighting, and prompts the official through required steps in order.
The agent suggests varied questions so that sessions do not follow a predictable script, which RBI expects to help establish that the interaction is live.
The official records the outcome. The agent stores the recording with date and time stamp, check results and a structured audit pack.
Sessions are queued for concurrent audit with risk flags, so auditors review high-risk sessions first.
It improves productivity, compliance, fraud control and customer experience.
The agent works with the lender's V-CIP platform and onboarding stack.
It connects to the video platform through APIs or SDKs to access streams, controls and recordings.
Aadhaar e-KYC or offline XML, PAN verification and CKYC registry integration for existing records.
Liveness, face match and document verification, including KYC document verification.
Status and outcomes flow back to the onboarding journey or LOS so the application can move forward, for example into an instant personal loan decisioning flow.
Recordings and evidence go to secure storage with retention and access controls.
Suspicious sessions feed fraud case management.
Lenders typically measure productivity, success rates, compliance quality and fraud detection.
More completed sessions per official per hour.
Higher first-attempt completion and fewer drop-offs.
Shorter average handling time per session.
Fewer observations in concurrent and internal audits.
More impersonation and replay attempts stopped before account opening.
The agent supports V-CIP across products and customer types.
Full KYC for borrowers onboarded through apps and partner journeys.
Converting accounts opened through Aadhaar OTP e-KYC, which carry limits, to full KYC.
Remote KYC for card and point-of-sale credit customers.
Periodic KYC updation for higher-risk customers through V-CIP.
Remote KYC for co-applicants who cannot visit a branch.
Fresh identity confirmation when fraud signals appear.
It gives onboarding and risk leaders better data on process performance and fraud.
Leaders see exactly where and why sessions fail.
Session demand by hour and day helps staff officials efficiently.
Linked signals reveal organised fraud attempts.
Official-level quality metrics support coaching.
Product teams compare V-CIP outcomes across sources and partners.
The agent must strengthen, not weaken, compliance with V-CIP rules.
The official must remain in control. Automation must not replace required human steps.
The V-CIP infrastructure must meet RBI expectations on security, encryption and data storage. Lenders should test the agent against these requirements.
Liveness and face match models can produce false results. Thresholds need tuning and human review for borderline cases.
Models should work fairly across skin tones, ages and disabilities. Alternate paths are needed for customers who cannot complete standard checks.
Video recordings and biometric data are sensitive. Consent, purpose limits and retention rules must follow the DPDP Act and KYC rules, and Aadhaar masking must be applied where needed.
Attackers adapt. Detection models need regular updates.
The agent is designed around the requirements in RBI's KYC Master Direction. Lenders should always check the latest version.
The session must be live and real time, conducted by a specially trained official of the regulated entity.
Geo-tagging must confirm the customer is physically in India.
The official must confirm liveness and match the customer with officially valid documents, and capture the customer's photo and PAN.
Sessions must be recorded and stored securely with date and time stamp.
Accounts opened through V-CIP should become operational only after concurrent audit.
If a session is disrupted, it should be restarted rather than continued from a break, according to the lender's procedure.
A phased rollout reduces risk and builds confidence.
Document every V-CIP step and the evidence each step creates.
Begin with customer guidance, step prompts and audit pack creation.
Introduce liveness, face match and document checks with conservative thresholds.
Have compliance and concurrent auditors review outputs during the pilot.
Expand to all officials and track productivity, success rates and audit results weekly.
Video KYC will become faster, smarter and more resistant to fraud.
Detection of synthetic faces and injection attacks will improve rapidly.
Agents will predict demand and match customers with available officials in real time.
Guidance will be available in more Indian languages and accessible formats.
KYC, credit decision, agreement execution and mandate setup will run in one continuous flow.
Industry and regulatory guidance on AI use in KYC will become more detailed.
It is an AI system that supports the bank or NBFC official who conducts a Video based Customer Identification Process, or V-CIP. It prepares and guides the customer before and during the call, runs liveness, face match and document checks in real time, captures geo-location, timestamps and recordings, prompts the official through each required step and assembles a complete audit pack, so officials complete more compliant sessions per hour.
No. Under RBI's KYC Master Direction, V-CIP must be carried out by an official of the regulated entity who is specially trained for the purpose, through a live, real-time interaction. RBI permits technology such as AI and face matching to support the integrity of the process, but the official remains responsible for the identification and the decision.
Typical checks include liveness detection, face match between the live customer and the photo on Aadhaar or other officially valid documents, PAN verification, document authenticity and tampering checks, geo-location to confirm the customer is in India, IP and VPN checks, detection of other people or prompting in the frame and confirmation that Aadhaar offline XML or e-KYC data is recent and valid.
Many video KYC calls fail because of poor connectivity, bad lighting, missing documents or customers who are unprepared. The agent runs pre-call checks on network, camera and documents, guides customers to good positions and lighting, schedules calls at suitable times and helps reconnect or reschedule interrupted sessions, which raises completion rates.
For every session it stores the recording with date and time stamp, geo-tag, IP address, liveness and face match scores, document images, the questions asked and answers given, official identity and the final outcome. These records support concurrent audit and inspections, and they are retained under the lender's record retention policy.
RBI requires accounts opened through V-CIP to be operational only after concurrent audit to ensure the process was followed. The agent prepares a structured audit pack for each session and highlights any steps with low scores or gaps, so auditors can review faster and focus on genuinely risky sessions.
It uses liveness detection, injection attack detection, face match against official documents, behaviour signals and randomised questions to detect impersonation, replayed videos and synthetic faces. Suspicious sessions are flagged to the official and to fraud teams, though no system can guarantee detection of every advanced attack.
Lenders typically see more completed V-CIP sessions per official per hour, higher first-attempt success rates, shorter average call times, fewer audit observations, faster concurrent audit and better detection of impersonation attempts.
Guide customers, automate liveness and document checks and build audit-ready records for every V-CIP call. Talk to our AI experts about a Video KYC Session Assistant AI Agent for your onboarding team.
Ahmedabad
B-714, K P Epitome, near Dav International School, Makarba, Ahmedabad, Gujarat 380051
+91 99747 29554
Mumbai
C-20, G Block, WeWork, Enam Sambhav, Bandra-Kurla Complex, Mumbai, Maharashtra 400051
+91 99747 29554
Stockholm
Bäverbäcksgränd 10 12462 Bandhagen, Stockholm, Sweden.
+46 72789 9039

Lewes
16192 Coastal Highway, Lewes, Delaware 19958, USA