Four patents. One system no one else can legally build.

A hardware company with proven results but no defensibility is a company a better-funded competitor copies. The deepest objection an investor carries into this category is the same one every time: what stops someone bigger from doing this. The moat answers it before the traction page even raises it. The Gaming results land harder behind a wall than in front of one, because the reader meets them already knowing they cannot be replicated.

The four patents are walked below as four layers of one wall, each with its number, its issue date, and the precise thing it blocks. Then the second wall, the data set. Then the reason a competitor cannot qualify on the six axes that define the category. Each patent number here is verified against the issued filing on file before the page asserts coverage.

Read as one wall, the four patents are not four separate filings that happen to belong to the same company. They are four layers that each force a would-be competitor to fail at a different point. Clear the hardware and the system patent still catches you. Clear both and the physics has no workaround. The expansion patent extends the same logic into the next vertical before anyone else arrives.

flowchart TB
    E["Entrant wants to build a real-time free-pour system"]
    E --> L1{"Layer 1: Hardware
US 10,769,589"}
    L1 -- "build the mat" --> BLOCK1["Infringes the
sensing array + RFID antenna"]
    L1 -- "design around" --> L2{"Layer 2: Full system
US 11,537,986"}
    L2 -- "mat to cloud to AI to app" --> BLOCK2["Infringes the
end-to-end architecture"]
    L2 -- "design around" --> L3{"Layer 3: Physics
US 11,715,064"}
    L3 -- "RFID near liquids" --> BLOCK3["No known workaround"]
    L3 --> L4["Layer 4: Expansion
US 11,983,670
the next vertical is already claimed"]

The patents stop a copy of the machine. The transaction data set stops a copy of the intelligence. Even a competitor who somehow licensed around every patent would still face a cloud AI trained on a pour-level transaction history they cannot quickly reproduce. To rebuild it they would have to install thousands of mats and wait years collecting real pours, all while infringing the patents that make the mat possible in the first place.

The mechanism is a flywheel that compounds rather than degrades. Every new venue feeds the same model, so accuracy improves for every venue at once. The data set is the asset a well-funded entrant cannot buy off a shelf, because it does not exist anywhere else. It is earned one pour at a time, and Klynkz started earning it years ago.

flowchart LR
    V["More venues install the mat"] --> D["More pour-level
transactions captured"]
    D --> A["The AI gets more accurate
for every venue at once"]
    A --> M["Higher margins,
stickier customers"]
    M --> V
    G["A new entrant starts here
with zero history"] -.-> D

The two transaction figures above are marked for source verification and route to the claims register before they anchor a data-moat claim in any data room. Diana rates the data moat as emerging rather than established at the current install base, and the page carries it that way: the asset is real and unique if the counts hold, and it matures from emerging to established exactly as the install base scales. That maturation is part of what the raise funds.

The moat is not only legal. There is a functional white space no competitor occupies. The category has six axes that matter to an operator, and Klynkz is the only system that clears all six at once. The real incumbent is not a rival product. It is the manual weekly count, still run for free in most independent bars, and it fails the same axes.

WISK and BevSpot are separate competing apps, not a merged entity. Both are software-first: they still require a human to capture data by photo or entry, which is why they sit at partial on the axes that depend on automation and real-time capture. Flow meters cover only metered lines and alter service. Smart scales need a person to place each bottle. Only the last column clears every row.

Put the two walls together and the conclusion is forced rather than asserted. A competitor who wanted to take this position would have to clear four issued patents, solve a physics problem with no published workaround that is not already owned, and then spend years installing mats to collect a pour-level data set that does not exist anywhere else, all before reaching the accuracy Klynkz already has. The cost and the time are the moat, and they compound against the entrant every quarter Klynkz keeps deploying.

Everything the rest of the section proves, the Gaming and MLB results, the recurring economics, the market, lands behind this wall. The traction page is next, and it reads differently once you know it cannot be replicated.