Lexius Turns Your Existing Security Cameras Into AI Guards That Actually Catch Shoplifters

The Macro: Retail Theft Is a $100 Billion Problem With Stone Age Solutions

Retail shrinkage cost the US economy over $100 billion last year. That number includes internal theft, administrative errors, and organized retail crime. The industry response has been a mix of hiring more security guards, installing more cameras, and hoping for the best. The cameras record everything. Nobody watches the footage.

I mean that literally. Most retail security cameras are recording devices, not monitoring devices. The footage sits on a hard drive until someone needs it for an insurance claim or a police report. By the time anyone reviews it, the shoplifter is long gone, the slip-and-fall has already turned into a lawsuit, and the store has absorbed the loss.

There are existing computer vision security products. Verkada sells smart cameras with analytics. Deep Sentinel offers live monitoring. But most of these require ripping out your existing camera infrastructure and starting fresh. That is expensive, disruptive, and slow. The average retailer already has cameras. They just need those cameras to be smarter.

The Micro: No New Hardware, Just Smarter Software

Lexius works with your existing cameras. Hikvision, Axis, Dahua, Hanwha, whatever you already have installed. The setup takes under 10 minutes and happens remotely. No on-site installation, no new hardware, no expensive upgrades.

Once connected, Lexius runs real-time detection for shoplifting, slip-and-fall incidents, and known offenders. The shoplifting detection sends instant video alerts. The known offender recognition works across an entire chain, meaning if someone is caught stealing at one 7-Eleven, every other location in the network gets alerted when that person walks in. That chain-wide intelligence layer is genuinely interesting and hard to replicate without scale.

David Elskamp and Liam Webster founded the company. David is a 2x founder who built a six-figure media agency in high school. Liam has a UC Berkeley EECS degree and did AI and privacy research at ICSI. They have a three-person team from YC Winter 2026. The client list is impressive for a company this young. 7-Eleven, Erewhon, and Prada are not easy logos to land, and they suggest the product actually works in production environments.

The pricing comes in three tiers based on camera count, all including free remote installation. An Erewhon case study claims $3,000 in monthly savings, and another client reports 60+ hours saved per week. Those are real, measurable ROI numbers that make the sales conversation straightforward.

The Verdict

Lexius has a clean product-market fit. The problem is obvious, the solution is elegant, and the go-to-market is frictionless. “Keep your cameras, add our software, stop losing money” is an easy pitch for any retail operations manager.

The competitive pressure will come from two directions. First, camera manufacturers like Verkada and Motorola will build these features natively. Second, cloud-based security analytics platforms like Ambient.ai are going after similar use cases. Lexius needs to stay ahead on detection accuracy and build enough chain-wide data that the network effects become a real moat.

In 30 days, I want to see the false positive rate. If store managers are getting flooded with alerts for things that are not actually shoplifting, they will turn it off. In 60 days, the question is whether the known offender database is growing fast enough to provide real value across chains. In 90 days, I want to see geographic expansion. This product works globally. The question is how fast they can scale sales without a large team.