From side mirrors to transmissions, the countless parts pouring out of a single end-of-life vehicle are a vast "gold mine" in themselves. But until now, this market has been thoroughly analog.
A startup has appeared in the roughly 200 trillion won global auto recycling market, which long relied on greasy ledgers and workers' memories, outfitting it with a "digital engine" of artificial intelligence (AI) and big data. That company is Amass, led by CEO Park Won-jae. Amass is a solution corporations that identifies and verifies auto parts based on the vehicle identification number (VIN), and it is aiming to expand into the global used auto parts transaction market worth about 116 trillion won.
In a recent interview with ChosunBiz, Park described the used auto parts market as "a market of unlicensed experts." Even ordering a single front bumper for a BMW is tricky because the name varies by model and model year. Without three to five years of experience, it is hard to place an order correctly even for one part. On top of that, once purchased, returns are almost impossible in this market. The moment a headlamp costing more than 10 million won is ordered incorrectly, the loss falls squarely on the repair shop or the parts distributor.
In fact, it is estimated that about 70% of the parts generated from end-of-life vehicles in Korea are distributed without accurate part identification information. Amass estimates that the scale of forgone transactions or disposal at low prices caused by this reaches about 2.5 trillion won per year.
Amass's "PartsFit" was developed to solve this problem. By entering only the VIN, users can instantly check data on about 1.43 billion items, including part numbers, compatibility, and diagrams. Amass currently holds data for 14 brands and more than 126,000 models.
Productivity gains have also been confirmed on the supplier side at dismantlers. One dismantler that handles 300 vehicles a month cut the part registration time per vehicle from 30 minutes to 5 minutes after adopting PartsFit. It also reduced the staff for part registration from 15 to 2.
Amass plans to extend its solution from parts management into distribution. Park believes dismantlers can be transformed into "high-margin manufacturers that disassemble cars into parts and sell them."
In Europe, used parts are priced at about 70% of new parts. In Korea, it is only 30%. Amass calculates that exporting Korean used auto parts at retail to global markets such as Europe and the United States would completely change the margin structure.
In expanding the global distribution market, the metric Park especially emphasizes is the "return rate." For example, the average return rate in the U.S. offline auto parts market is about 22%. As vehicles become more complex, more parts look the same on the outside but differ in the number of pins or internal programming.
While local corporations maintain the practice of "if it's wrong, just allow a return," Amass presents as a key competitive edge that it can lower the return rate to around 6%. That is the average return rate in the U.S. online auto parts market.
In global markets, the strategy is to supply the solution in API form instead of software as a service (SaaS). The judgment is that directly integrating into existing platforms is more expense-efficient.
Park also expressed confidence about the spread of electric vehicles (EVs). "Just because it's an EV doesn't mean the process or difficulty of building data changes," Park said. "In fact, in the EV market, where part unit prices are high and there is a strong need to verify repair costs, the value of Amass's solution will grow." Park noted this also aligns with the potential to expand into the global insurance repair cost verification market.
More than half of Amass's current revenue comes from the domestic insurance-exclusive solution "CA." It has the same basic structure as PartsFit, but in the process of repairing damaged vehicles, AI verifies in real time whether the claimed parts are appropriate and whether the quantities match. This used to rely on the experience and skill of insurance adjusters.
Park said, "Until now, even if repair shops overclaimed to meet the '2 million won surcharge threshold,' there was little justification to screen it out, but at one client, the rate of erroneous claims that was 24 out of 100 cases at the start of Amass's adoption has now fallen to the 6% range." The interpretation is that the market's self-correction has begun simply because a verification system is in place.
DB Insurance reduced losses by about 4.65 billion won in 2025 through Amass's solution for the year. The ROI (return on investment) compared to the solution usage expense is about threefold. The time required to process loss adjustments for damaged vehicles was also shortened from 49.9 minutes per case to 8.7 minutes, resulting in about a 5.7-fold improvement in work efficiency.
Park said, "Ultimately, the goal is to present a 'standard plan' that everyone can accept transparently for repair cost claims." Related measures are currently being discussed with the General Insurance Association of Korea.
For entry into the global insurance market, the primary leverage will be cooperation with clients. "We are discussing ways to leverage DB Insurance's existing network," Park said, adding, "Independently, we have also completed proofs of concept (PoCs) with State Farm and Progressive, the largest insurers in North America."
Amass has raised a cumulative 9 billion won. It is targeting an initial public offering (IPO) in 2028. "Through 2028, expanding market share in the domestic insurance market is the top priority," Park said. "After completing Europe and North America data development this month, we plan to begin full-fledged global expansion starting in 2028."