A Swiss Train Maker Is Putting LLMs in Front of Its Utah Foremen

On 22 October, Stadler US posted a role it calls an AI Enablement & Improvement Lead for its Salt Lake City plant. The job ad opens with an ambition few rail manufacturers have put in writing: an "AI-enabled industrial organization" where human expertise and AI work together on decisions, productivity and quality. The hire will train office employees, functional leaders, shop-floor team leads and foremen to use AI, especially LLMs, and then prototype and scale solutions with approved tools.

Two details make the posting unusual. The ad states that the job is "not a pure data-science, software-development, or IT support role" and describes it as business transformation. And the only tool it names is Microsoft Copilot, alongside enterprise systems such as SAP, Microsoft 365, Jira, Primavera and PLM. Stadler's own careers site files the vacancy under purchasing.

Martin Ritter runs Stadler North America; Philipp Brunner becomes group CEO in January.

A plant growing faster than its playbook

Stadler US is a decade old and growing quickly. It became a standalone division of the Swiss group in January 2025 under Martin Ritter, who sits on the Group Management Board. A factory expansion opened in July, doubling the manufacturing area to 44,100 square metres and promising up to 300 more local jobs on top of 674 people already employed in the area. The plant builds car bodies, bogies and wiring harnesses for customers including Utah Transit Authority, MARTA in Atlanta and Caltrain.

The group has more work than it can easily absorb. Stadler's half-year results showed revenue up 40% to CHF 1.97 billion and an order backlog of CHF 33.3 billion, roughly six times this year's revenue guidance. The EBIT margin of 4% sits below the group's 6% to 8% target for the year. Philipp Brunner will take over as group CEO from Markus Bernsteiner on 1 January 2027, with Peter Spuhler remaining chairman. A US plant that can lift output per head without matching hires one for one becomes valuable in that picture.

Stadler's backlog is about six times its expected 2026 revenue.

Desk tools on a welding floor

Here is where the plan will be tested. Copilot lives inside Outlook, Teams, Word and Excel, and most foremen spend their shifts beside welding cells, assembly jigs and harness boards rather than inboxes. Rail cars are built to customer specifications, federal safety rules and Buy America requirements, and their designs belong to a Swiss parent that guards its engineering. An LLM summarising a non-conformance report or drafting a work instruction touches exactly the documents that auditors read.

The ad anticipates some of this. It asks for an understanding of data privacy, intellectual property, cybersecurity and access control, and it requires the hire to coordinate complex use cases with IT and Stadler Group functions. Keeping the role outside IT means it can move faster, though it also means the person must persuade IT to grant access rather than own it.

The case for starting with people

The counter-argument is strong. BCG's AI at Work survey found frontline use of generative AI stuck at 51% while leaders sat at 85%, with usage rising sharply where people received hands-on, in-person coaching. Microsoft's 2026 Work Trend Index found 65% of employees feel pressure to adopt AI, yet only 13% feel incentivised to innovate with it. A trainer who knows the shop floor addresses both gaps directly.

Most employees feel pushed toward AI; few feel rewarded for experimenting. An animated version is in the asset pack.

Stadler also has form here. Its Utah plant co-founded a youth apprenticeship that pays students to earn an advanced manufacturing degree while they work, and it trained local welders by bringing in Swiss staff. Peers are moving the same way. Deployed's manufacturing desk has tracked how Honeywell took generative AI to 95,000 employees, and Walmart has argued that AI fails when people are left out. Just days earlier, Safariland's first AI hire was briefed to train process owners face to face. AIM's coverage of learning and development leaders points the same way, as does a Kyndryl finding that only a quarter of organisations feel workforce-ready.

The AI role arrives between a factory expansion and a change of group CEO.

Five questions for Martin Ritter

Q1Which shop-floor tasks will foremen use LLMs for first?
Q2Do foremen have Copilot licences and devices on the floor today?
Q3How will AI-assisted documents be reviewed before they reach customers or auditors?
Q4What can the US plant build locally, and what needs sign-off from Stadler Group?
Q5How will Stadler measure the value of the role after its first year?

Stadler built its US business by teaching Utah workers how Swiss trains are made. Its new hire will try to teach the same workforce how to work alongside AI. If foremen on the welding floor are using LLMs by the time Brunner takes charge in Bussnang, Salt Lake City may end up teaching the rest of the group.