RAMP AI at McConway & Torley:
From Guesswork to Answers

McConway & Torley has cast the couplers, knuckles, and coupler components that have held America's rail network together since 1869. Its Pittsburgh foundry remains one of only four coupler producers left in the world. The company continues to invest in its equipment, including the automated coremaking line at the center of this case study.

But like most complex manufacturing floors, the equipment runs on PLC logic with misleading alarms and outdated documentation. The fastest path through a fault has always run through whichever experienced workers happen to be on shift. For M&T's crew operating and maintaining the coremaking machine, this meant the team had access to diagnostic information — but what they really needed was a faster way to make sense of it.

Premier Labs brought RAMP AI onto the coremaking line to close that gap, deploying it in two modes. Seventeen maintenance technicians across all three shifts, ranging from one month to seventeen years of tenure, used Ask & Explore — RAMP AI's conversational mode trained directly on M&T's own PLC code, wiring diagrams, pneumatic schematics, and maintenance manuals. They used it to ask questions in plain language and get answers grounded in the machine's actual documentation. A smaller group of key employees also used Guided PLC Diagnostics, RAMP AI's mode for working a live fault end-to-end.

Faster Answers, Fewer Guesses

During the engagement, technicians used RAMP AI to work through real faults on the coremaking line — not simulations.

Those results held up beyond the two events above. Across the technicians who completed scored assessments during the engagement, 69% estimated a diagnostic speed improvement of 50% or greater, and 94% said RAMP AI gave them enough information to avoid escalating a fault entirely.

What Stood Out Beyond the Stopwatch

The clearest signal came from the most experienced troubleshooter in the group — a maintenance supervisor with seventeen years on the floor and the person other technicians already turn to when they're stuck. Working through a real recurring fault, he judged RAMP AI's reasoning "spot on" and confirmed it matched the exact troubleshooting method his own team already used.

Technicians also flagged RAMP AI's ability to reconcile the different internal terms people use for the same part. One maintenance lead watched RAMP AI's answer correct itself once he clarified a component's name after using the team's informal term for it — a common problem when two technicians on the same floor rarely describe the same part the same way.

Two additional benefits emerged clearly:

1. New hires stand to gain the most. Newer technicians were especially eager to use RAMP AI, since it gives them a way to troubleshoot with more independence, earlier in their careers. That's especially valuable on the third shift, where fewer senior staff are typically on hand. A new employee moving to third shift said he's looking forward to having RAMP AI to help him work through problems on his own.
2. Compounding value. Because RAMP AI records maintenance events for each machine, the value compounds over time. Technicians on the same shift described being able to see what a colleague already tried in a prior session — turning a shift change into a handoff rather than a guessing game.

Where This Goes Next

The results at M&T reflect a straightforward pattern: RAMP AI reasons over the PLC logic, alarms, and documentation a facility already has, and returns plain-language guidance that holds up against the judgment of the most experienced person on site.

Premier Labs and M&T are now moving from this engagement toward a broader deployment across additional machines on the Loramendi line.