See What Others Can't: Why We Built BevMatrx AI

Sunday night

It is 11:40 on a Sunday and the bar is finally quiet. The last table left an hour ago. The floor is mopped. And Dana, who runs the beverage program for three venues, is standing in a walk-in with a clipboard, doing inventory.

Dana is not a real person. She is a composite of a dozen people we have worked beside, and if you have run a bar, you already know her. She has been doing this inventory for six years. She knows the back bar by feel. She can tell you the velocity of every well spirit without looking. She is very good at her job.

On Monday morning the owner calls. Beverage cost for the month came in at 24 percent. The target is 21. "Where did the three points go?"

Dana does not know. Not because she is careless. Because nobody could know. The three points are hiding somewhere in the gap between what the invoices say she bought, what Sunday night's inventory says is left, and what the register says was sold. Three systems. Three exports. One person with a spreadsheet, and a Monday that already has forty other problems in it.

So she gives the honest answer every operator has given at least once: "I'll look into it." And she will. She will pull the POS report, check the comps, ask the bartenders about the Saturday rush, and by Thursday she will have a theory. By then the month is half over, the three points are still leaving, and the next inventory is Sunday.

Here is what that Monday costs. One venue doing $900,000 a year in beverage. Three points of cost nobody can explain is $27,000 a year, leaving quietly, with no one to blame and nothing to fix. Now multiply by every venue Dana runs, and then by every Dana in the group. (Illustrative math. Your numbers are the ones that matter, and the first thing we do is get them.)

The beverage program is usually the highest-margin line in the building. It is also the only line where "I'll look into it" is an acceptable answer to "where did the money go." That is the problem we built a company to end.

Who we are

We have been Dana. That is the whole reason BevMatrx AI exists.

My co-founder, Jean-Pierre Torrealba, spent 25 years running multi-unit hospitality operations, most recently as Director of Operations for a group of more than forty locations. He has opened more than a dozen restaurants and bars. He has stood in more walk-ins on more Sunday nights than anyone should, and he has sat across from more owners asking where the three points went than he could ever tally. He is the reason this product thinks like an operator instead of a spreadsheet.

After completing my M.S. in Data Science at Carnegie Mellon I took my first venture-backed startup from concept to scale and have since overseen the opening of over half a dozen hospitality concepts. I did not plan for those three things to collide, but they did, in a walk-in, holding an invoice in one hand and an inventory sheet in the other, realizing the answer to the question plaguing Dana was sitting right there in the data. Nobody had built the thing that reads it.

Our CTO, Donald Steinert, has spent the last decade building restaurant technology. He has been the first technical hire at three startups, took one from first commit through acquisition, and has shipped production systems into kitchens and bars that cannot afford downtime on a Friday night. He is the reason the product works at 11:40 on a Sunday and not just in a demo.

Operators, a founder who builds, and an engineer who has lived in this industry. We are based in Miami with our roots in New York hospitality, and we are building BevMatrx AI alongside our first design partners, which means the operators using it today are deciding what it becomes.

What we are solving

Dana's three points are not one problem. They are two, and they have different stories. Most tools blur them together, which is why the three points never get found.

The money that leaves

This is the story of the three points. Every beverage program has a gap between what should have sold, based on what you bought and what is left after inventory, and what the register says sold. That gap is variance, and it is where margin leaks. An over-pour that became the house pour. A recipe that drifted half an ounce. Comps that were approved with a nod and never rung in. A case of wine that turned. Bottles that died in an ice bucket at 2 a.m.

None of these is dramatic on its own. Together they are the three points. And because they arrive as one blended number at month end, most operators do the only thing they can: absorb it. It becomes a line item called "unaccounted for," and a line item is where money goes to disappear quietly.

Here is the question we think every owner should be asking, and almost none are. Beverage net sales is beverage sales minus comps. Net sales looks fine. But why is $3,000 a month leaving before it ever becomes net sales, who approved it, and how does that compare with last year? Comps are the single most auditable piece of beverage loss, and in most groups nobody audits them, because the number sits inside a total and the total looks fine.

We believe variance is a question, not a verdict. When you can see it by product, by period and by cause, you stop absorbing it and start fixing it. And the fixes are rarely dramatic. A pour spec. A comp policy with a name on it. A recipe card that matches what the bar actually makes. Each one holds for months. The money was always there. Now you can see it.

The money that sits

This is a different story, with a different villain. Walk into Dana's storage room and you will find it: four cases of a mezcal from two menus ago. A wall of a vodka the bar was already long on. A premium gin that was somebody's idea in March. None of it is missing. All of it is paid for. It is cash that already left the bank and is now sitting on a shelf, earning nothing, waiting to become a write-off.

It got there honestly. Ordering runs on par levels, and par levels were set once, under a menu that no longer exists, by a manager who no longer works there. So the sheet says "order," and the storage room fills with product that sells a case and a half a week. Dead stock is the same mistake further along: product that stopped moving and never got a reason to move.

Dead stock is not a storage problem. It is a balance sheet problem. The money is already spent. The only question left is whether it comes back as net sales or leaves as a write-off. Stop buying what you already own.

And if you run more than one venue, you are watching both stories through a keyhole

A director of operations with twenty venues sees twenty spreadsheets, twenty inventory habits and twenty versions of the truth. A CFO sees one line. The bar manager who knows exactly what is wrong has no way to show it. So the same mistake gets made in ten places at once, the venue that is quietly outperforming never gets copied, and the one that is bleeding never gets caught in time. More locations. Less guesswork.

The value we deliver

Picture Dana's Monday a few weeks in. The owner calls. Beverage cost came in at 24 against a 21 target. "Where did the three points go?"

This time she has the answer before he finishes the question. One point is a well tequila pouring heavy since the new bartender started on the 12th; the pour spec went back on the rail this morning. One point is comps, up 40 percent at the Friday bar, almost all approved by one manager; there is a conversation at 3 p.m. The third point is a case of rosé that turned in the walk-in, written off and already pulled from the next order. And by the way, the storage room is carrying $11,000 of product that has not moved in ninety days; here is the special that sells through the mezcal by Thanksgiving. (Illustrative, of course. But this is what the conversation sounds like when the three points have names.)

That conversation is the product. Everything else is how we get you there. Run a beverage program on evidence instead of guesswork and it produces the same three outcomes, every venue, every week, and they are the only three we care about:

• More revenue from smarter sales. The product running low is in the well on Friday, and the product that was not moving gets a reason to move. Dead stock becomes net sales instead of a write-off.
• Lower costs. Buy what sells, not what the spreadsheet says, and close the gap between the bottle and the register. Variance gets a cause, the cause gets fixed, and the fix holds.
• A more profitable operation, with the receipts to show for it.

How we do it

Your beverage program already contains every answer Dana needed. The evidence lives in three places you already have: your invoices (what you bought and what you paid), your inventory (what is on the shelf) and your point of sale (what actually sold). Almost nobody puts the three together, because they live in three systems and the only thing connecting them is a person with a spreadsheet at midnight.

BevMatrx AI connects to the systems you already run, whatever point of sale you use, and reads invoices, inventory and sales as one record. From that record it produces the thing a spreadsheet never could: a recommendation with the reasoning attached, ranked by what it is worth in dollars, so the biggest number is always at the top of the list.

• Variance you can act on. After each inventory, what you should have sold against what you did, by product and by period, so margin leak gets a name instead of a feeling. The three points stop being a line item and become a list of things to fix.
• Ordering on data, not on a par column. You are covered for nine weeks on this SKU. Skip the next order. Here is the working capital it frees. And when a product is running down, it says so early enough to matter.
• Dead stock, turned into net sales. What stopped moving, what it cost you, and what it would take to sell it through before it becomes a write-off. Feature it. Run it as the special. Put it in a cocktail that is already moving.
• One clear picture for every level. The bar manager, the beverage director, the director of operations and the CFO see the same numbers, with the detail each of them needs. Every bottle. Every dollar. Every decision.

What we will promise is this: the first thing BevMatrx AI shows you is your own inventory, read back to you, with the money in it made visible. Most operators see something they did not expect within the first week. Usually it is a number they had been absorbing for years.

How to work with us

It is built for groups that run more than one venue: restaurant groups, nightlife, hotels with multiple outlets. If you run one bar, start free on a single location and grow from there.

Get in touch with us below to set up your restaurant.

See what your inventory is missing

Dana deserves a better Monday. So does the owner. So do you.

See What Others Can't.
Every Bottle. Every Dollar. Every Decision.