First of all, I don’t think any of us are sitting around waiting for PSA, BGS, or CGC to suddenly roll out “AI grading” or tampering detection. People grade with whichever company they want to grade with for their own reasons, and if you like TAG for the reasons you already stated, that’s perfectly valid.
I’m only giving you my perspective. For long-term value, I don’t see TAG as the best option for grading.
I’ve been pretty vocal about TAG because I think they market themselves in a way that makes people think they have some incredible technological moat, when they really don’t. A lot of what they present as revolutionary is just a polished version of computer vision workflows that are very achievable for any serious grading company with capital, volume, and imaging infrastructure.
Using high-resolution imaging, segmentation, surface analysis, defect detection, and classification models to assist grading is completely doable. That part is not magic. In fact, it is the opposite of that. It also does not inherently require some groundbreaking leap in AI/ML. In a lot of cases, the hard part is not “building AI,” it’s deciding what standards you want the system to enforce, then operationalizing that and handling the transition period with employees as they learn to use new tools.
That’s why I’ve repeatedly said on E4 that the AI side of TAG’s business is not some massive barrier to entry. If anything, the larger grading companies are in a much stronger long-term position to build similar or better systems because they already have scale, scans and data for training, some of the necessary imaging infrastructure already in place, historical data, and massive submission volume. Those things matter a lot more than slapping “AI grading” on marketing materials or paying influencers to crack 10s from other companies and heap praise on the “AI grading process.”
And as restoration, trimming, cleaning, and other forms of tampering become more common, companies like Collectors are only going to have more incentive to improve their own detection workflows anyway, and that would not necessarily require AI. So I don’t really buy the idea that TAG is permanently out in front here in some untouchable way. They’re actually incredibly vulnerable and seem to be hoping for acquisition.
My bigger issue with them is the marketing. Things like the self-grading machine concept sound great in theory and probably sound great to investors too, but in practice that kind of thing becomes messy very fast. It’s the same pattern: present something futuristic, let people assume the technology is more advanced or defensible than it really is, and hope that perception carries the brand.
So my point was never that TAG cannot detect useful things, or that no one should grade with them. My point is that the “AI grading” angle is doing a lot of heavy lifting in their branding, and I think a lot of collectors are mistaking that branding for a real moat.
