Skip to content
Fast Company Executive Board

The Fast Company Executive Board is a private, fee-based network of influential leaders, experts, executives, and entrepreneurs who share their insights with our audience.

What video games taught us

How to get factory workers to actually use AI.

[Photo: Getty Images]

Profile photo of Hugo Dozois-Caouette

CTO and Co-founder · MaintainX

I’ve watched a maintenance technician stare at a brand-new database dashboard for 10 minutes without clicking anything. He was smart, but the software gave him 47 options and no obvious starting point.

We build tools that assume users will power through complexity with a good-enough return on investment. Users won’t do so. And in today’s manufacturing environment, the patience threshold is lower than ever, with higher employee turnover, shorter job tenures, and the near-disappearance of the 20-year “power user.”

The solution, surprisingly, comes from game design.

In role-playing games like Elden Ring, you don’t see the weapon upgrade system until you’ve survived your first few encounters. You start simple—move, attack, dodge—and the game reveals depth as you prove you’re ready for it.

Game designers call this “progressive disclosure,” and it’s one of the most effective onboarding techniques ever developed. It also happens to be exactly what enterprise software gets wrong.

Day-one overload and solutions

Over the past few decades, traditional enterprise tool modules have overloaded factory team members, resulting in overlapping and confusing workflows and options.

The cost is staggering. A McKinsey study found that field technicians can waste up to 40% of their workday on non-value-adding activities, such as paperwork, idle time, and unnecessary travel, rather than doing their actual work. That’s a productivity crisis, and bloated software is a major contributor.

Technology can instead offer tools that feel intuitive, learn alongside users, and drive higher adoption—and better data.

For industrial software, progressive disclosure looks like a new technician logging in and seeing one thing: Their work orders. That’s it. They learn to complete and close work orders.

As confidence builds, the system displays the preventive maintenance schedule. Then parts inventory. Then, meter-based scheduling and real-time industrial automation systems, with each layer unlocked as proficiency grows.

Compare that to the traditional approach: You enter a system and face a highly complex tool. Dozens of workflows and modules are available at once, difficult for someone to onboard and understand what to do first.

Making this work is more achievable now than ever. Improved feature flagging and code tooling made phased rollouts practical at scale.

Strike a balance

Early features must deliver immediate value without frustrating your existing users, who know more advanced capabilities exist.

We offer practical patterns: expandable forms, “show more” controls, guided onboarding tours, and explicit signaling that tells users what’s coming next. When done well, these yield faster onboarding, lower training burden, and higher completion rates.

Progressive disclosure solves visual overwhelm by controlling what a user sees and when. But it can’t touch the second layer of complexity, of the work itself.

Even after you’ve simplified the interface, a technician must write a proper work order, request parts, and troubleshoot a machine they’ve never worked on before. AI then comes in as progressive disclosure’s natural counterpart.

Progressive disclosure simplifies what you see, while AI simplifies what you do. A user describes what they need in plain language, and the AI executes.

AI also asks clarifying questions, interacts with the underlying system, and learns preferences over time. Right now, few manufacturers provide real-time visibility into production data. It’s a massive win to connect team members to that data through a conversational interface handling behind-the-scenes complexity.

AI changes the work

These workflows now run in production, making front-line workers’ lives easier.

Consider the shift handover. A shift supervisor once spent 30–45 minutes at the end of every shift compiling a handover report, pulling from action logs, summarizing what happened, and flagging what’s still open. Now, AI autogenerates that report from the shift’s actions and history. The supervisor reviews, revises, and hands off a clean summary in minutes.

Or take the Monday-morning scramble. When someone calls in sick, a scheduler reassigns work orders to match needed skills and adjusts priorities based on estimated times and production demands. Now, AI handles the combinatorial logic, while the planner approves the plan.

Troubleshooting is where it gets interesting. AI can synthesize machine manuals and historical work orders into conversational, step-by-step, real-time guidance. A technician can say, “I tried that—now what?” and get a context-aware next step. Every interaction improves the knowledge base.

In inventory and procurement, AI compares vendors, generates purchase orders, and suggests optimal parts with agent-style automation, which required a dedicated analyst several years ago.

AI also provides critical background intelligence by pooling anonymized knowledge across dozens of company sites to standardize troubleshooting while preserving data privacy.

Where AI fits

AI works best as an accelerator for repeatable, high-stakes processes, such as financials, inventory transactions, and compliance workflows. AI can also unlock lightweight automations and can “see” actionable patterns in immense amounts of data that no human ever could.

But AI also requires clear boundaries, human-in-the-loop approval, and transparent compliance documentation to counter legitimate managerial fears. We take governance seriously because trust is the prerequisite for adoption.

What does this all mean?

AI’s broader promise is in reclaiming wasted time, making software resilient to turnover by capturing tribal knowledge, and turning software tools into an “experienced colleague” available 24/7.

But I keep returning tothe norm when the interface disappears, and the factory’s unique business logic remains its competitive edge.

So, sell the Tuesday morning when a technician walks in, sees what needs to be done, asks one question in plain language, gets a useful answer, and starts turning wrenches.

Progressive disclosure gets users in the door. AI keeps them effective once they’re inside. Together, they replace the old model, the 300-page user manual, the launch-day overload, and the assumption that everyone will eventually figure it out.

Industrial software will soon make every feature feel like it was there when you needed it, and make the hard parts feel easy when you arrive. That’s how AI adoption stops being a scary leap and becomes a natural evolution on the factory floor.

Hugo Dozois-Caouette is the chief technical officer and cofounder of MaintainX.