Pilot
See it run in your venue
Stock never quite reconciles. You suspect leakage but cannot prove it, cannot locate it, and do not want to accuse a team you depend on. A pilot is how you find out what your own bar actually does — a number from your room, in your week, rather than a case study from someone else’s venue.
What it is for
One camera over the pour zone, one box in the building, and an independent count of every drink poured — reconciled against your till, shift by shift and station by station.
Where the two records agree, you learn your operation is clean. Most weeks, on most stations, that is what happens — and it is worth knowing. Where they diverge in a pattern, you finally know when, where and on which SKU.
Nothing leaves your building except counts. All inference runs on a node in the venue, there is no facial identification in the stack, and every event is attributed to a station rather than to a person.
It starts with a survey
We look at the room before anything is installed, and we tell you whether it is a good candidate.
The pour zone, the camera positions actually available to you, your stations and your volume. That is enough to say whether the signal will be there.
Some rooms are not candidates. Cocktail-led venues with heavy multi-ingredient builds and low spirit-by-peg volume are a poor fit. So are rooms where no overhead camera position exists. So is a bar doing too little volume for the variance to clear the noise inside a pilot window. We would rather say so at the survey than after the install.
On numbers
We publish our error floor and nothing we have not measured.
Accuracy against a manual count, recovered variance and payback all come out of the pilot itself — reported with the conditions they were measured under, including the shifts where the model was weak. Until then we would rather show you the measurement than an estimate of it.
What runs in the room
Barvis-X1 Vision
The flagship model. A camera over the pour zone produces its own count of every drink poured — at peg granularity, with a confidence band.
Full specificationReconciliation Engine
Matches the vision ledger against the POS ledger and surfaces the divergence. Source-agnostic by design.
Full specificationBarvis Edge Node
One box in the venue. All real-time inference runs on premises; raw video never leaves the building.
Full specificationTell us about your room
The number of pour stations, which POS you run, and whether cameras are already installed. That is enough for us to say whether a survey is worth your time.