62% of summit respondents pick training and digital work instructions

Lincoln International's experts attended the Connected Worker Manufacturing Summit in October 2026, where the audience was polled on investment priorities. Sixty-two percent picked workforce training and skills management, or digital work instructions, as their top connected-worker investment priority for the next 12 months. Lincoln describes the results as directional.

According to Lincoln International's experts, connected worker investment is shifting. The focus used to be digitizing frontline workflows and surfacing operational insights. Now buyers want solutions that produce measurable operating outcomes.

Three numbers from the summit floor

Lincoln calls the results directional, which is fair for audience polling at a conference, but the pattern holds together.

Asked what would most help their operation today, 56% of respondents pointed to reduced downtime or improved labor productivity. If buyers expect the payoff there, a connected-worker business case that rests on feature counts or adoption rates will likely face more resistance than one framed in minutes of downtime and labor hours.

For 29% of respondents, frontline decision support was where AI could deliver the most operational value.

Decision support is easier to judge with a concrete picture in hand. SymphonyAI's glossary defines a connected worker as someone who can see real-time equipment or process performance data, pull up procedures and engineering drawings, and use an AI copilot to diagnose or troubleshoot problems in the field. LightGuide, in a July 2026 post, separates artificial intelligence, where systems generate insights and recommendations, from what it calls augmented intelligence, where people get intelligent tools at the point of execution.

Read that way, the 29% and the 62% are describing the same worker. A copilot's recommendation only helps if the instruction behind it is current and the person receiving it is ready to act.

From paper replacement to closed-loop action

Lincoln frames the market as three stacked phases. The first, digitize, covers systems of record: paper gets replaced, data gets captured and processes become standard. The second, connect, produces systems of insight by letting people, assets and systems share real-time visibility and collaborate. The third, execute and optimize, is where Lincoln expects spending to go. In that phase, decisions are AI-supported, action is coordinated and results can be measured where the work actually happens.

Vendor material shows what the first phase looks like on the floor. SymphonyAI describes digital work instructions as the move from paper to interactive, step-by-step guidance with real-time updates, sold alongside digital inspections and skills training modules backed by skills matrices that track employee development. The same glossary also pitches persona-based industrial AI copilots that help with team scheduling and task management based on skill levels. In Lincoln's terms, that's a third-phase promise running on first-phase data.

Lincoln's first trend explains why that third phase matters. Manufacturers want to close the loop between spotting an issue and carrying out the response, and they want to cut the decision latency that builds up across manual handoffs and disconnected systems. Put plainly, it's the gap between a system that reports a problem and one that gets the right instruction to the person who has to fix it.

That gap depends on more than software. A system can only route a task to a qualified person if it knows who's qualified. That's why the training result fits the execution thesis instead of contradicting it.

Competency becomes the data that runs the floor

Lincoln's third trend states the pressure plainly: retirements, labor shortages and turnover are making it more urgent to capture tribal knowledge before it walks out the door. Lincoln also sees manufacturers shifting from checking that training was completed to confirming that workers are actually competent and ready. Job and skills data, kept under tight governance, is more and more often tied to scheduling, machine access, workforce planning and career progression.

Once skills data decides who gets scheduled on a line or cleared to run a machine, a training-completion checkbox stops being good enough. That reasoning sits behind the 62% figure. Training and work instructions aren't a side project next to AI. They supply the data an execution system needs before it can assign anything to anyone.

Other research points the same way as the summit poll. LightGuide cites Deloitte's 2025 Smart Manufacturing and Operations Survey, where 35% of respondents said adapting workers to the "Factory of the Future" ranked among their top concerns. Part of that concern is giving workers the skills and tools they need to get the most out of smart manufacturing. The two surveys asked different questions of different groups, so their figures can't be added together. Still, each one places workforce readiness near the top of the agenda.

The same pressure shows up in process industries. Writing in Hydrocarbon Processing in December 2025, CJ Renegar of EFESO Management Consultants argued that the retirement of seasoned refining and petrochemical operators has left newer, less experienced employees without the knowledge to spot improvement and cost-saving opportunities. His remedy leans on standard work, which he defines as the current best, safest and most efficient way to do a job. Simplifying, codifying and, where possible, digitizing those standards can boost productivity, improve cost predictability and speed up onboarding during the skills shortage, he writes.

Renegar also ties frontline capability to the outcomes the summit audience cared about. Engaged frontline staff who make real-time decisions while operations are underway improve reliability and control costs, he argues, through less overtime and fewer repairs. That suggests the 62% training priority and the 56% downtime-and-productivity expectation are two ends of one business case rather than separate wish lists.

Where embedded AI earns a spot at the station

Lincoln lists the frontline jobs AI is moving into: creating work instructions, retrieving knowledge, troubleshooting, making recommendations and guiding execution. The ROI case it describes rests on less administrative work, faster decisions and workers who perform better in the field and on the shop floor. Lincoln adds that trusted data, governance and human oversight remain critical to adoption.

The case for placing that AI at the workstation begins with who actually does the work. LightGuide cites an industry survey showing that people perform 72% of factory tasks. It also points to Deloitte research in which just 23% of frontline workers reported having the tools required to be productive. Taken together, the two numbers show that most of the work happens where most workers say their tools come up short.

For manufacturers whose quality and throughput still depend mostly on manual assembly, inspection or maintenance, equipping those workers is a more concrete place to look for AI value than another analytics layer.

The strongest objection comes from Lincoln's fourth trend: deployment doesn't produce business impact on its own, sustained frontline adoption does. Lincoln lists usability, workflow integration, executive sponsorship and local champions as essential to changing daily behavior, and says leading solutions replace legacy workflows rather than adding another disconnected tool.

Academic fieldwork shows what that objection looks like up close. Carnegie Mellon researchers Esther Kang and Sarah Fox spent 20 months studying an airport's pilot of autonomous UV floor-cleaning robots deployed during the COVID-19 pandemic, published at the DIS '22 conference. The robots couldn't act on their own and needed careful calibration and coordination from janitorial staff, one of whom said tracking them added ten more minutes to the routine. Workers also raised safety concerns in congested areas and asked for more comprehensive training across all levels of the organization.

An airport cleaning crew isn't a production line, and floor robots aren't work-instruction software. The pattern still maps onto Lincoln's fourth trend. A tool that adds a task instead of retiring one looks a lot like the "another disconnected tool" Lincoln warns about, and the people closest to the work notice first.

Plant-wide rollouts and the systems behind them

Lincoln's fifth trend may matter most to the person who writes the purchase order. When they buy, manufacturers are considering global deployments instead of isolated pilots. Integrations with ERP, EAM, MES and HCM systems are gaining importance, along with common standards and data governance. Buyers also have to weigh global consistency against local flexibility on languages, regulations and workforce needs.

Vendor feature lists show how that integration checklist gets answered in practice. SymphonyAI, for example, advertises connections to SharePoint, MES, LIMS and SAP ERP systems, an open API and LTI-enabled learning management integration, plus an offline-first mode in which operators keep working without connectivity and data syncs once they reconnect. For a multi-site buyer, the learning-system link is where training records would meet the scheduling and machine-access decisions Lincoln describes, and offline operation could matter at sites where connectivity varies.

Lincoln's sixth trend sums up the buying posture: manufacturers are buying outcomes rather than software categories and putting measurable improvement first.

The signal to watch over the next year is how the 62% spend their training and work-instruction budgets. Bought as standalone tools, they stay in the systems-of-record phase. Bought as the skills layer connected to scheduling, machine access and the plant's ERP and MES, they become the foundation Lincoln's execution phase needs, and the downtime and labor-productivity numbers will show which way each plant went.