How HotPass AI Is Different From the AI Already in This Market
Let's not pretend AI in weld inspection is a new idea. It isn't, and any reader who's shopped for AUT or RT equipment in the last few years already knows it.
Every major AI offering in this space is bolted to hardware you have to buy from the same vendor. Rhythm runs on Waygate's own detectors. WeldXprt is Eddyfi's AUT ecosystem. PipeWIZARD is Evident's phased-array system. If your shop's current fleet is a mix of legacy film scanners, a Waygate CR unit from one project, and a different AUT rig from another — which describes a lot of Canadian NDT shops running equipment bought in different capital cycles from different vendors — none of these AI layers travel with you. You get the AI by replacing the hardware, not by pointing it at what you already own.
Baker Hughes' Waygate Technologies ships Rhythm software with FLASH! processing alongside its CRxVision and DXR detectors — deep learning trained on millions of inspection images to auto-classify defect types and severity. Eddyfi Technologies has been building AI into its next-generation AUT software tools, including WeldXprt for pipeline girth weld inspection. Evident's PipeWIZARD iX does phased-array AUT on girth welds and heat-affected zones with its own analysis layer. And the direction isn't speculative — Saudi Aramco's Zuluf project validated an AI-driven automatic defect recognition system for weld radiography at 95% accuracy, cutting read time to under a minute.
Industry analysts peg the AI-assisted NDT analytics market at roughly $520 million in 2026, growing toward $1.46 billion by 2036, with radiography/CT alone accounting for close to a third of it.
So the honest question isn't "does AI work for weld inspection." That's already answered. The question is what you're actually being asked to buy to get it, and what happens after the read.
That's the same lock-in problem Enrisk built Trucision to solve on the OEM side of the lab-instrument world: "no firmware rewrite" instead of forcing a hardware refresh. HotPass AI applies the same logic here. It reads what comes off whatever you're already running — film scans, CR, DR, or real-time radiography — without asking you to standardize on one vendor's imaging stack first.
And every one of those tools stops at the read. Rhythm, WeldXprt, and PipeWIZARD are built to classify the defect in the image. None of them are in the business of building the reporting pipeline, the chain-of-custody record, or the CWB/CER-ready audit package around that classification — that's left to whatever spreadsheet or PDF workflow your shop already has, which for most of the market is still manual. HotPass AI's parent, Enrisk, is a build shop, not just a software vendor — the same engagement that puts an AI read in front of your team can also automate the paperwork that currently eats the time the AI read just saved.
To be direct about it: if your shop already has a Waygate, Eddyfi, or Evident system in place and it's doing what you need on the read side, that's a real, working solution — this isn't a claim that the incumbents don't work. The gap is narrower and more specific: hardware-agnostic AI for shops running mixed or legacy equipment, plus the workflow automation layer none of the equipment vendors are selling at all.
Where this goes next: if you're evaluating AI for weld inspection and the pitch you've gotten so far starts with "which of our imaging systems do you want to buy," that's worth a second conversation. HotPass AI starts with what's already on your bench.