The AI Inflection Point

In January 2024, after 98 years of publication, Modern Machine Shop published its first artificial intelligence cover feature. The conversation then was exploratory — ChatGPT writing G code, researchers asking whether AI could be trusted with expensive machines and materials. The conversation today is different. AI is quoting jobs in small machine shops across the country. It is sanding ambulance panels in Iowa without a single line of custom code. It is adjusting feeds and speeds across an entire CAM program in response to a voice command. And this September, for the first time in the show's history, IMTS will give it a room of its own. 

According to data from All About AI, 43% of manufacturers are actively implementing AI — in predictive maintenance, quality control, process optimization, production planning, and knowledge capture. That number will be visible on the show floor at McCormick Place. But before we get to Chicago, it's worth understanding what problem AI is actually solving, and why the urgency is real. 

THE PROBLEM: SPEED AND LABOR 

For U.S. machine shops, the competitive pressure is intense. "Proximity should be an advantage," says Alex Huckstepp, co-founder and CCO of Uptool, an AI quoting platform built for small and mid-sized shops. But it often isn't. Buyers who want to work with local suppliers are getting same-day quotes from overseas and parts in a week. Meanwhile, domestic shops that are spending 15 to 20 minutes quoting each part manually are losing work not on price, but on response time. Uptool's AI connects to a shop's quoting inbox, parses drawings and bills of material automatically, and gets that time down to roughly 90 seconds per part. As co-founder Benny Buller says: "The biggest opportunity in the manufacturing of parts is not in developing the next modality of manufacturing technology. It is in software that allows better flow of information and faster decisions."  

Award-winning (RBR50) Scan&Sand solution at Life Line Emergency Vehicles, sanding sections of the vehicle which previously were very hard for operators to process. Now with the solution, operators are able to make the robot take on the brunt of the work, freeing them up for more valuable tasks around the shop.

The labor problem is just as real. Every Friday at 3 p.m., Ryan Steffen, paint supervisor at Life Line Emergency Vehicles in Sumner, Iowa, walked the production line looking for volunteers to sand ambulance panels. Nobody wanted the work. GrayMatter Robotics' AI-powered Scan&Sand system changed that — scanning each vehicle's unique geometry and beginning work without custom programming, cutting sanding time by more than 30%. "It doesn't take away jobs," Steffen says. "It handles monotonous, taxing tasks." 

These two problems — front-office speed and shop-floor labor — showcase the range of where AI is producing measurable results today. 

THE SPECTRUM: FROM INBOX TO ROBOT CELL 

AI in manufacturing is not a singular technology. It is a spectrum from software that processes email to robots that learn the physics of grinding a new alloy. 

At the front-office end, tools like Uptool and ECI Software Solutions' AI Foundry are making the case that ERP infrastructure is the prerequisite — without centralized, structured data, AI has nothing to learn from. Jim Belosic, founder and CEO of SendCutSend, who has built more than a million lines of custom shop management software, is direct about the current limits of AI-assisted development: "It can put out a lot of spaghetti and a lot of slop. Maybe in the future it can just be AI but today is not that day." Justin Gray of Toolpath Labs made the same point at a recent industry summit: your process has to be stable before the next layer of software or automation works. 

In CAM programming, the changes are moving faster. Mastercam's Copilot — embedded in Mastercam 2026 — allows programmers to adjust feeds and speeds across an entire program, create machine groups with multiple operations, and search the company's entire video tutorial library, all via voice or text command. Russ Bukowski, Mastercam's president, frames the technology carefully: "We still have so much knowledge tied up in experienced programmers. What we're doing is offloading the tasks we do have reliable information for, so the user can focus on the complex problems that AI can't solve yet." Open Mind Technologies' Hypermill Data Center takes a complementary approach — on-premise machine learning that searches a shop's own past projects, keeps customer data local for IP protection, and typically saves 70 to 80% of programming time on automated portions. 

An AI-powered robotic grinding system performs a weld-blending operation at GrayMatter Robotics’ California facility. The system uses force control and 3D scanning to adapt in real time to part geometry and material variation. Source: GrayMatter Robotics

At the far end of the spectrum is what GrayMatter Robotics co-founder Dr. Satyandra K. Gupta calls "embodied AI." Digital AI produces output a human reviews before anything physical happens. A 99% accuracy rate is great as long as the human catches the 1%. Physical AI executes actions in the world without that review step. On a 200-step robotic finishing process, a 99% accuracy rate means two errors per part — scrap or rework, every time. "If even one motion segment of a robot path is wrong and if the grinding wheel crashes into the part, it's a disaster," Gupta says. GrayMatter's Scan&Grind system — which scans parts, self-programs motion, and monitors force at kilohertz frequency — is engineered for error rates closer to one in a million. Trener Robotics approaches the same high-mix challenge differently, using natural language instructions to generate and execute production plans for CNC cells in real time, eliminating the need for custom programming entirely. 

THE HONEST COUNTERWEIGHT 

Smaller shops face three specific sticking points when evaluating AI: fear that it will replace skilled workers, distrust of non-deterministic behavior with expensive machines and materials, and legal ambiguity around IP ownership of AI-assisted output [LINK: "CAM Copilots and the Next Digital Shift in American Machining," Donaldson/MMS]. At a recent industry summit, Justin Gray described his own guardrail: "There is no way to stop an LLM from doing dangerous stuff. The one hard line for me is that I never give the LLM control over Git, so I can always roll back whatever it did." [LINK: "Before You Automate, Learn to Stop," Donaldson/MMS] The underlying principle applies beyond code: AI amplifies what's already there. Stable processes get more stable. Unstable ones fail faster. 

WHERE IMTS 2026 COMES IN 

AMT's Emerging Technology Center Source: AMT

 

For the first time in the show's history, IMTS is giving AI its own dedicated floor space. The Industrial AI Arena — 25-plus exhibitors at the front of North Hall — focuses exclusively on applied AI for the factory floor. Production-ready tools, not concept demonstrations: quality inspection, predictive maintenance, process optimization, demand forecasting, cybersecurity, and safety — all in one environment. 

The companion Industrial AI Conference (Wednesday, September 16, South Building S401, $450 advance registration) was developed with Dr. Jay Lee of the University of Maryland Center for Industrial AI. Lee has identified the three barriers manufacturers most commonly face: finding usable data, selecting the right tools, and deciding whether intelligence should run at the edge or in the cloud. The full-day program addresses all three with sector-specific case studies and a roundtable on implementation trade-offs. 

Beyond the Arena, AI runs through the entire show. The Software sector includes Mastercam, Toolpath, CloudNC, Open Mind, and Uptool, all exhibiting. The Automation sector — 260-plus exhibitors, the largest it has ever been — features natural language programming platforms and zero-CapEx automation models that lower the entry point significantly. In the South Building, AMT's Emerging Technology Center will operate a live manufacturing system — producing drone airframes in partnership with Oak Ridge National Laboratory, Mazak, and Fanuc for all six show days — demonstrating how digital twins, AI, and automation interact under real production constraints. 

Michael Bauer, owner of Marathon Precision in Wheeling, Illinois, frames the show's value proposition this way: "For the week of the show, IMTS is the largest, most advanced shop in the world. If you buy technology before anybody else and get good at it, chances are you can win new business."  

With 43% of manufacturers already implementing AI and a dedicated arena at the Western Hemisphere's largest manufacturing technology show, the question of whether AI belongs in manufacturing is settled. The question is where your shop starts. Bukowski's view on where this leaves the professionals running these shops: "They'll be the ones adapting AI for their own use. Long term, we'll see a shift in what these professionals do — from executing routines to creating them. The value of their expertise is only going to go up."