StradVision Patent Portfolio Analysis

Eighty-eight million, arriving in August 2022

One event, one number. StradVision announced an $88M Series C on 1 August 2022. For a South Korean company working in artificial intelligence, that is the moment the outside world agrees something has been built. It is the natural place to start reading a file, because most portfolios are shaped by the money: capital lands, headcount grows, filings follow.

This one reads backwards. The company's first filing was made on 11 May 2015, more than seven years before the raise. By the time the Series C was announced, the file already held the overwhelming majority of what it holds today. The money did not start the work. It arrived after the work had been done, and the dates are unambiguous about the order.

One hundred sixty-six granted, five still pending

Split the 171 families by status and the shape is lopsided in a way you rarely see. One hundred and sixty-six are granted. Five are published and pending. That is a 97 percent grant rate across five patent offices: the United States, Europe, China, Japan and South Korea. Examiners in all of them have finished with almost everything here.

A pending stack is a company's forward edge. It shows what is still being argued, still being narrowed, still unresolved. Here that edge is almost gone. What remains is a settled body of rights rather than a work in progress. The narrow ribbon is not a failure of throughput. It is the signature of a filing programme that was front-loaded and then closed.

The pending stack is where a company's future arguments live. This one has five.

Eighty-nine families in a single year

The bars tell the plot. Two families in 2015, four in 2016, seventeen in 2017, twenty-five in 2018. Then 2019: eighty-nine families, more than half the entire portfolio filed in twelve months. Twenty-three in 2020. Six in 2021. Nothing recorded for 2022, the year of the raise. Two in 2023, three in 2024.

The funding marker sits in the trough, not the peak. Allow for the shaded window, since filings from the last eighteen months may not be published yet and the most recent published filing is dated 26 March 2025. Even so, the 2019 spike is three years clear of the money. Whatever conviction produced eighty-nine families, it was acted on long before anyone validated it with a cheque.

Three names on almost every family

Ninety-one inventors are named across the file. Three of them carry it. JE HONGMO appears on 155 families. Kim Yongjoong on 154. RYU WOOJU on 153. Against a total of 171, that means each of those three is named on roughly nine of every ten families the company holds, including the eighty-nine filed in 2019.

The remaining eighty-eight names cluster around them. This is not a distributed research organisation with parallel programmes; it is a small core with a very high output rate and a large supporting cast. The four most recent published titles cover automatic labeling of training data, pseudo 3D bounding boxes, and annotation based on lidar maps. The same core carried the file into that territory.

One hundred seventy-one against several thousand

Set the file against everyone else filing in the same CPC groups over the last five years and the scale changes. UNIV ZHEJIANG has 3,057 families there. SAMSUNG ELECTRONICS CO LTD has 2,642. TENCENT TECH SHENZHEN CO LTD 2,491, STATE GRID CORP CHINA 2,178, UNIV XIDIAN 1,877, UNIV TSINGHUA 1,731. StradVision is the small circle: 171 families, most of them filed in one year.

Size is not the comparison that matters here. Those groups are where image recognition, machine learning and data processing claims all land, so they are crowded by definition. What the record shows is that a three-person core took a dated position in that crowd in 2019, saw it through to grant in five offices, and left almost nothing unexamined behind it.

Density, not volume, is what this file was built on.

What the dates say and the pitch cannot

Here is the thread pulled straight. StradVision's patent record is not a story about capital being deployed. It is a record of a small group betting early, filing hard in 2019, and then largely stopping. The $88M arrived after the bet had already been placed and mostly examined. That is the rarer and stronger sequence, because the priority dates cannot be bought later.

It also means the file has a defined edge. Five pending families, three years since the peak, and a core of three inventors who wrote almost all of it. Everything that follows now happens outside what has already been granted. The question is not whether the position was taken. The dates settle that. The question is what comes next.