Four stages, each ending with something your loss prevention team can use and measure, not a report that sits in a shared drive.





A chain-wide LP suite flags what is typical for everyone. Your losses cluster in specific stores, shifts and SKUs. We encode the rules your LP team already trusts as auditable code, add models only where patterns are too subtle for rules, and keep review screens on low-code.
We haven't published an in-store loss prevention case yet. The closest work: return-fraud risk scoring for a DTC retailer that cut fraudulent returns 38% and protected $180K in margin (representative engagement, anonymized under NDA), plus a computer-vision returns verifier piloting at a beauty retailer. Explore our case studies.

Transaction and employee records stay inside your private cloud. Nothing trains a public model, and access to investigation data is role-based and logged from the very first prototype.
You own the code, rules and models. Opening a new location never adds a subscription line, and your IT team can extend the system without asking us for permission.
Every alert shows the transactions, rules and signals behind it, so an investigator can defend the decision to a store manager or in an employment review.
Start with one store cluster or one fraud type. The same architecture extends to new locations, channels and data sources without a rebuild as your retail loss prevention program matures.
$15,000+
A focused prototype that tackles one loss driver, such as refund abuse, void patterns or inventory gaps, on your real data in about two weeks. Best for independent retailers and small chains testing whether custom AI beats their current reports.
What's included?
Shrink Audit: ranked map of loss drivers across stores, SKUs and transaction types
Prototype Model: exception detection running on your real POS or returns data
Data Connectors: read-only pipeline from one POS or e-commerce platform
Findings Report: measured impact, false positives and the fastest path to production
$38,000+
A production loss prevention system with risk scoring, investigator case queues and shrink dashboards across your store network. This is the tier most mid-market retailers choose once the prototype proves where margin is leaking.
What's included?
Everything in Shrink Prototype: hardened for daily use and deployed in your private cloud
Risk Scoring: explainable scores for refunds, voids, discounts and employee transactions
Case Management: low-code investigator queue with evidence attached to every flag
Integration: secure connection to your POS, ERP or inventory management system
$75,000+
A full-scope loss prevention platform spanning stores, e-commerce returns, receiving and vendor billing, with agentic workflows that triage cases automatically. Built for multi-location retailers replacing a patchwork of point tools.
What's included?
Everything in LP Analytics Platform: extended across all channels, stores and loss types
Multi-System Integration: POS, ERP, WMS, e-commerce and video management system event feeds
Agentic Triage: AI agents that assemble evidence and draft case summaries for investigators
Compliance & Training: SOC2/GDPR-aligned deployment plus hands-on onboarding for your LP team
Enterprise LP suites assume hundreds of stores and a dedicated analytics team. A regional retailer needs something smaller, sharper and owned outright.
Policy and void rules run as transparent code, and machine learning is added only where patterns are too subtle for rules.
Case queues show evidence, history and reasons, so your LP team spends time on real cases instead of false alarms.
A working detector runs on your own transactions within 14 days, before you commit budget to a full build.
You own everything we build, so adding stores, registers or investigators never raises a license fee.
The work runs in four funded stages, and you can stop after any of them and keep everything built so far, from the audit findings and data model to the prototype code itself.
Step 1
We review your POS, inventory and returns data and pinpoint the two or three loss drivers most worth automating first, ranked by margin impact.
Step 2
A working detection model goes live on your real transactions within about 14 days, with results you can measure against today.
Step 3
We connect POS, ERP and case management, add guardrails and deploy everything inside your private cloud with full audit trails.
Step 4
We compare shrink, false positives and investigator hours to your baseline, then extend the system to new stores and loss types.
Loss happens wherever goods, refunds and invoices change hands. We apply the same detection engine across the retail value chain, from the checkout lane and returns desk to the receiving dock and the AP queue.
Every engagement is led by a founder. These are the people you will work with, and the experience they bring to retail loss prevention.



Our engineers average 5–7 years of senior experience, and the founding team has run fraud prevention systems serving millions of users.
Answers from our team to the questions retailers ask before trusting software with loss prevention: which signals matter, how much data is enough and where rules beat AI.
Most retailers know their shrink rate but not its source. Send us a few months of POS and inventory data, and a solutions architect, not a salesperson, will show you where the losses cluster.
A working prototype runs on your real data in about 14 days. The production timeline depends on how many data sources and stores are involved, and we commit to a dated plan after the audit. For reference, our comparable return-fraud system went from kickoff to live in about four weeks.
Projects start at $15,000+ for a focused shrink prototype. A production LP analytics platform starts at $38,000+, and an omnichannel build at $75,000+. Unlike per-store SaaS licenses, you pay for retail loss prevention software you own, and we size scope against the shrink you can realistically recover.
Yes. You own the repository, trained models, data pipelines and documentation outright. No proprietary platform or per-store license holds your retail loss prevention system hostage, so your own team or another vendor can extend it after launch without starting over.
Everything runs inside your private cloud perimeter under SOC2/GDPR-aligned practices and a signed DPA. Your data never trains public models. Risk scores use behavioral and transaction signals only, never protected attributes, and every flag is logged, explainable and confirmed by a person before any action.
Yes. We build around your current stack instead of replacing it. Typical sources include POS transaction logs, ERP and inventory adjustments, e-commerce and returns platforms, and event feeds from video management systems that expose an API. We applied the same pattern in a license plate recognition system built on an existing camera network.
We measure shrink, confirmed-fraud rate, false positives and investigator hours against your pre-launch baseline, using your own cost inputs. After handover, you can run the system in-house, keep us on for iteration, or expand to new loss types. There is no mandatory retainer.