Cash Counter Agent
Reconciles every till transaction against camera evidence — catches skip-scan, ring-down and no-sale-open patterns. Flags suspicious behaviour for human review.
“We find the shortfall at month-end — long after the cash walked.”The riskiest three feet in the store, watched every second.
Everyone learns to live with the gap between “something happened” and “someone noticed.” We did not. Let the cameras do the watching and your people do the deciding — alerted the instant it matters, not hours later. From “that is just how it works” to “wait — it could work like this.”
Most retailers accept till-side shrinkage as a write-off. Industry data says they shouldn't.
Industry surveys consistently report retail shrinkage as a measurable percentage of revenue, with internal till-side incidents (skip-scan, ring-down, no-sale-open, refund fraud, sweetheart deals) making up a significant share. The size of the loss at your specific stores is benchmarked during the pilot.
CCTV alone does not catch this. Even with a camera pointed at every till, no security manager has time to watch every transaction. Loss-prevention teams catch a fraction of incidents in arrears, weeks after the cash is gone, by which point the cashier has rotated and the evidence is academic.
The result is a write-off that everyone accepts as normal — a small percentage every month, never investigated because the per-incident value is below the investigation cost. It is only when you total it across 50 stores across a year that the number gets ugly.
Every till. Every transaction. Every drawer-open. Reconciled live — and surfaced for human review.
Cash Counter Agent watches the cash drawer, scanner zone, and customer interaction at every till simultaneously. Per the VIZO361 datasheet (Use Case 12 — Cash Theft Identification), it monitors counter / transaction areas at 4 MP minimum cameras with focused positioning, with approximately 3 days of per-till deployment plus POS-integration time.
Each detected event — drawer open, hand entering scan zone, item moved across scanner, customer hand exchange — is timestamped and matched against the POS transaction stream in real time. Mismatches surface as exceptions for loss-prevention review. The agent does not require the cashier to do anything different — it runs on the cameras you already have, with the POS feed you already publish.
Critically, Cash Counter does not confirm theft. Per the datasheet, it flags suspicious behaviour patterns for human review. Every flagged event requires loss-prevention sign-off before any HR or disciplinary action. False-positives are reviewed and used to improve site-specific tuning during the pilot phase.
Known limitations (from the VIZO361 datasheet): the agent cannot confirm theft — it only flags suspicious behaviour for review. False-positives are expected and tuned per site during the pilot. Customer faces are blurred by default in stored evidence.
How Cash Counter Agent works — end-to-end
From the moment a frame leaves your camera to the moment your hardware reacts. No black box — every step explained.
Define till zones
Draw zones on each till camera — drawer, scanner, payment surface, customer hand-exchange. Tag each with the POS terminal ID.
Detect till events
Vision model detects drawer-open, scan-pass, hand-into-drawer, refund-event, item-bagging — each with timestamp and confidence.
Sync with POS stream
Live transaction feed from POS (LS Retail, Posist, Ginesys, Square, custom) joins on terminal ID + time.
Reconcile + flag
Drawer-open with no transaction? Scan-pass without barcode beep? Refund with no return item visible? Each mismatch flagged with clip + transaction + cashier.
Daily exception report
Single-page report per store, sorted by financial exposure. Manager WhatsApp on shift-end if cashier-level pattern crosses threshold.
Detection without action is just an alarm. We close the loop.
Every Cash Counter Agent event can trigger your existing on-site hardware. Wired or wireless. PLC, contactor, BACnet, MQTT or webhook — your choice.
Manager Silent Alert
Integration
WhatsApp Business API, SMS, Slack — pushed to store manager / loss-prevention lead.
Outcome
Significant pattern (e.g. multiple no-sale-opens in a short window) → silent alert with clip + cashier identification. Manager observes discreetly.
POS-Hold Flag (optional)
Integration
REST API to POS — flags transaction for supervisor approval before close.
Outcome
High-confidence skip-scan event → POS holds transaction close; supervisor must approve. Used at high-shrinkage stores during shift transitions.
Loss-Prevention Dashboard
Integration
Web dashboard + nightly email summary to LP team / head of stores.
Outcome
Per-store / per-cashier / per-shift exception summary with drilldown to clip.
ERP / Inventory Push
Integration
Webhook to ERP (SAP, Oracle, Microsoft Dynamics, custom) for inventory adjustment on confirmed shrinkage.
Outcome
Confirmed loss events feed directly to inventory reconciliation. Stock counts stop drifting from book value.
Audit + Compliance Log
Integration
Immutable log (append-only, hash-chained). Exportable for audit / legal proceedings.
Outcome
Evidence stream that meets internal HR-investigation, audit and legal-proceeding standards.
Where Cash Counter Agent is deployed
Specific industries, specific scenarios, specific outcomes — not generic marketing claims.
Retail — Apparel, Electronics, Value
POS shrinkage + refund-fraud detection
Tills monitored across multi-store chains. Refund-fraud and ring-down patterns surface during the pilot. Site-specific reduction in tagged shrinkage benchmarked in your pilot report.
QSR & Coffee Chains
Drawer-open + portion-pour audit
Fast pace makes manual monitoring impossible. Drawer-pop without ticket is the highest-frequency loss event. Portion-pour audits become camera-verifiable. Some customers add cup-count vs ticket-count reconciliation.
BFSI Branches
Teller drawer + cash-handling audit
Branch teller drawers monitored for compliance. Cash-deposit and withdrawal events reconciled against teller-system records. Used as evidence in routine RBI / SAMA audit.
Petrol Pumps & Service Stations
Pump-and-drive + cash-fuel audit
Unbilled pump-and-drive events automatically flagged (paired with ANPR Sentinel) with plate + dispenser + clip. Cash-only fuel transactions reconciled against dispenser meter readings.
Hospitality & Hotels
F&B outlet + bar till audit
Restaurant and bar tills have higher shrinkage exposure due to volume and complexity. Skip-pour events at bars are notoriously hard to catch — Cash Counter Agent makes them visible without one auditor per bar.
Multi-tenant Retail Parks & Malls
Common-area + tenant-cooperative monitoring
Mall operators offer Cash Counter as a tenant-cooperative service. Tenants opt in for shared cost.
Technical specifications
The numbers we publish in proposals. Detailed accuracy targets, latency budgets, and integration scope.
| Datasheet reference | Use Case 12 — Cash Theft Identification |
| Deployment timeline | Approximately 3 days per till, per the datasheet (plus POS-integration time) |
| Minimum camera resolution | 4 MP |
| Training required | Moderate — per-till camera-angle and POS-feed integration tuning |
| Camera positioning | Cameras focused on counters / transaction areas, per the datasheet |
| POS integrations | LS Retail, Posist, Ginesys, Square, Lightspeed, custom REST / file-drop |
| Compatible cameras | Standard RTSP / ONVIF IP cameras, 2 MP–8 MP |
| Scale | Up to 500+ concurrent camera streams in the enterprise tier |
| Privacy | Customer faces blurred in stored evidence by default. Cashier identification via uniform / badge / Identity Agent. |
| Audit trail | Append-only, hash-chained log for evidence-grade investigation use |
| Benchmark accuracy | Site-specific — measured and documented during the 30-day pilot |
Shrinkage you've been writing off becomes shrinkage you can investigate. Behaviour shifts once the team knows the system is live.
Direct shrinkage: a customer with material annual revenue is writing off a measurable percentage every year as shrinkage. The recoverable share — through behavioural change, investigation, and disciplinary action — is benchmarked during the pilot and shared in a signed report.
Behavioural change: even before formal action on any individual cashier, the awareness that transactions are being reconciled in real time changes behaviour. Customers report a reduction in flagged events within the first month — pure observer effect.
Insurance posture: with auditable till reconciliation in place, customers have discussed reductions in commercial-crime / fidelity policy premiums with their brokers. Specific reductions are insurer- and policy-dependent.
A real-world example
Customer name withheld per NDA. Full reference available on request.
Reference customer (named under NDA)
Proeffico has deployed Cash Counter across reference customer sites in retail, QSR, BFSI, and petrol-pump segments. Verified outcomes are available under NDA on request. Site-specific reductions in tagged shrinkage are benchmarked during your 30-day pilot and shared in a signed pilot report.
Pilot
Real numbers benchmarked on your site, in your signed pilot report
Cash Counter Agent vs the traditional way
Most facilities still run on the old playbook. Here's what changes.
The traditional way
- ✗LP team reviews CCTV in arrears — catches a fraction of events weeks later
- ✗Skip-scan and ring-down invisible without one auditor per till
- ✗Cashier rotation erases evidence trail — investigations stall
- ✗Shrinkage tolerated as 'cost of doing business'
- ✗Refund fraud invisible — no link between refund records and customer-bag contents
With Cash Counter Agent
- Every transaction reconciled live. Daily exception report on the manager's desk by morning.
- Camera + POS reconciliation runs continuously across every till. No additional staff.
- Per-cashier behavioural pattern visible early. Action possible before rotation makes evidence cold.
- Per-store / per-cashier / per-shift loss decomposition. Evidence-grade clips for HR / police action.
- Refund events matched against customer bag / item exchange. Mismatches flagged in real time.
Questions buyers ask before they sign
Do we need a new POS for this to work?
+
No. Cash Counter Agent works with your existing POS via the standard transaction feed. We support LS Retail, Posist, Ginesys, Square, Lightspeed, and dozens of custom REST or file-drop interfaces.
What does the cashier see / know?
+
Up to you. Most customers display a notice at each till — 'transactions are being monitored for accuracy' — for transparency and as a deterrent. Others run silently. The observer effect is real either way.
Is this legal in India?
+
Yes, with appropriate notice. Workplace CCTV for preventing financial loss is within scope under the IT Act, DPDP framework and Industrial Employment Standing Orders. We recommend an employee-policy update at deployment. Template wording provided.
What if our shrinkage is at receiving / warehouse, not at the till?
+
Cash Counter addresses till-level shrinkage specifically. For receiving / warehouse shrinkage, see the Footfall Analyst and Crowd Density agents. A combined deployment can address the full funnel.
How do we handle false-positives?
+
No detector is perfect. Per the datasheet, the agent flags suspicious behaviour — every flagged event has the clip attached for human review. Confirmed false-positives are fed back to retrain the model. Accuracy improves over the first 60–90 days as the agent learns your specific till setup.
Can it integrate with our HR / disciplinary system?
+
Yes. Confirmed-incident exports can flow to your HRMS as a documented incident with evidence attached. We do not automate disciplinary action — the system surfaces evidence, humans make decisions.
Will my customer's faces be stored?
+
Customer faces are blurred in stored evidence by default. Where customer identity is necessary for refund-fraud investigation, the relevant clip can be unblurred via an admin workflow with audit logging — typically used only in confirmed-incident escalation.
What does it cost?
+
VIZO361 is sold on a Private Offer model — pricing depends on till count, sites, POS integrations and reporting depth. A 30-day pilot at one store with 4–6 tills is included.
From the live VIZO361° platform — not mockups
Screenshots and alert clips captured from the deployed product. Customer identifiers have been blurred where present.
Who built this — and who reviewed it
Last reviewed: 2026-05-22Product Lead
Utkarsh Yadav
AI / ML Lead, VIZO361° (Facial Recognition + Liveness Detection)
Leads the facial-recognition, liveness-detection, and identity workflows in VIZO361. Designs the agent architectures behind Identity, Footfall Analyst, and PPE Compliance.
Product Lead
Swastik Vaish
AI Engineer, VIZO361° (Fire & Smoke, Cash Detection)
Leads the safety + loss-prevention model family in VIZO361 — including Fire Watch, Cash Counter, Guard Watcher, and Crowd Density. Owns model accuracy benchmarking during pilots.
Reviewed By
Saurabh Agarwal
Founder & CEO, Proeffico Solutions Pvt Ltd
Founded Proeffico in 2018. Reviews every product narrative for accuracy against the VIZO361 datasheet, customer reference behaviour, and Proeffico's published trust commitments.
Profile →All capability claims on this page are anchored to the VIZO361 Product Datasheet (2025). Site-specific accuracy, latency, and ROI figures are benchmarked during a 30-day pilot on your cameras and shared in a signed pilot report — not published as universal numbers.
Explore other Vision Agents
Private Offer · No published list price
See Cash Counter Agent on your cameras. 30-day pilot. NDA. Real conditions.
Two-camera pilot on your site. Real models. Real lighting. Real workflows. Quote within 5 working days of discovery.