Nº 001 · Origin
This started inside a real operation.
Opaxa was founded by operators who run more than 40 locations across 7 major U.S. airports, with over 750 people. Every location is its own brand, menu, lease and labor contract, and demand moves with the flight board. We built Opaxa inside our own operation, to solve our own problems first.
Nº 002 · The Hundreds of Decisions
A multi-location operation runs on hundreds of small decisions a day.
How many people for the first hour. Which spec to buy from which distributor at this week's price. Whether to extend the late shift when demand runs long. Which review to respond to first. How hard to push the new menu item. Whether to escalate the contract complaint or absorb it.
All of them compound into the difference between an operation that breaks even and one that provides for the people in it. The hidden math of every operation is the cumulative weight of every small decision the team did not have time to make well.
Opaxa was built for those decisions.
Nº 003 · The Missing Layer
Every multi-location operator runs on great systems. Nothing connects them.
The point of sale knows the sales. The payroll platform knows the labor. The back-office suite knows the invoices. The supplier portals know the cost. The inventory system knows what is on hand. The review platforms know the sentiment. Each one does its job well, but they live in separate logins, and nothing bridges them. So the operator stitches it all together in their head every morning, then makes the call by gut feel before the day begins.
The industry solved measurement a decade ago. The missing layer is action.
The operator's stack tells them what happened. It was never built to do anything about it. Opaxa does not replace any of those systems. It sits above all of them, reasons across everything they know at once, and acts where they could only report. The schedule is built on the forecast. The order is placed knowing this week's price. The invoice is checked against the contract. The margin stops bleeding between systems.
Nº 004 · The Moment
Three things changed at once.
Models can reason. What used to require a team of analysts and a week of work can now happen in a few seconds. The frontier keeps improving on a timescale of months, not years.
Models can act. Agentic AI has gone from research demo to product. It can now work across software systems, take real actions inside production environments, and stay inside the guardrails an operator sets for it.
Models can be taught. A general model can now be specialized on the texture of a single operation: your menu, your vendors, your brand voice, the way your managers actually talk to a vendor.
Generic intelligence became operator-specific intelligence.
Nº 005 · The Category
The action layer is a new category.
The action layer is the part of the stack that never existed: software that turns what the systems already know into decisions and acts on them inside the operator's rules. It starts in food and beverage, because that is where margins are thinnest and decisions are most frequent. It extends to any operation that runs on locations, labor and inventory.
The defensible asset is not the model.
Every company can buy the model. The asset is the operating context: the contracts, leases, recipes, rules and corrections that make an agent right on one specific operation. That context is only captured by deploying inside real operations, and it compounds with every day the agents run.
Nº 006 · What We Learned
Our own operation taught us three things.
Before Opaxa was a product, it was a tool we built for our own locations. Our managers still work with it, and most of what we know came from them.
Demand does not follow the calendar.
In our operation it follows the flight board: a wave of delays moves the rush, and a gate change moves it to another location. Every operation has its own version, from the weather to the event down the street. The signals that move demand are rarely in the point of sale, so Opaxa reads them where they live.
No two locations are alike.
The lease, the brand standards, the approved vendors and the labor agreement change from one location to the next. A model trained on the average fits none of them, so every agent learns each location on its own terms.
Managers trust what they can check.
A recommendation without its reasoning gets ignored, however good it is. So every recommendation arrives with the reasoning attached, every agent starts by suggesting, and it takes on more only when the operator decides its record has earned it.
Nº 007 · The Agentic Loop
Every action is scored. Every agent improves.
Opaxa is not a static rules engine. Every action it takes generates outcome data. The flagged invoice line is credited or it is not. The schedule produces a labor percentage against the actual day. The forecast lands close to actual sales or it misses. The manager accepts the recommendation or overrides it, and the override carries more weight than anything else the agent learns from.
Every decision teaches the next.
The agents learn from each other, too: the forecast shapes the schedule, the schedule shapes the order, and every result shapes the next forecast. The longer Opaxa works inside your operation, the sharper it gets: week over week, month over month, on data that is yours alone.
Nº 008 · Where It Goes
From better decisions to operations that run themselves.
Agents recommend.
Across demand, labor, menu, purchasing and cash, agents put the right decision in front of the right manager, with the reasoning attached. Managers decide. Every decision and every correction becomes part of the operation's record.
Agents act.
As agents earn it, operators promote them to act inside approved limits. Agents begin to read the documents that actually run an operation: leases, vendor contracts, labor agreements, equipment manuals. The routine gets handled. The exceptions get escalated.
Operations run themselves.
The operation runs itself between decisions, at every location, in every kind of business that runs on locations, labor and inventory, from restaurants to retail. The manager's job becomes the decisions only a person should make.
Nº 009 · The Return to Hospitality
Logistics belong to software. Hospitality belongs to people.
The promise of this technology is not just a better margin. It is that the people who run these operations get to go back to running them. Train the team. Walk the floor. Read the room. Greet the guest. Taste the new dish. Sit at the bar at the end of service and listen.
These are the things software cannot do and never should.
The reason we built Opaxa is so that the operators we know can do the part of the job that called them to it in the first place. So the guests they serve get the experience the operators want to deliver, every night.
Nº 010 · Contact
Build it with us.
If you run many locations and want to see Opaxa on your operation, or you are an investor who wants the materials behind this page, get in touch.
The founders read every submission. We respond within two business days.