The Factory Is Still Running on the Shaft

The Agent XBE Collective met for its second session on May 4, 2026. The members came with live workflows. They left with something harder to get on your own.

Before electricity, factories were organized around a single power source -- a steam engine driving a central shaft. Every machine had to be placed close to the shaft. Layouts were fixed and linear, dictated by mechanical power transmission. When electricity arrived in the 1880s, factory owners did the obvious thing: they replaced the steam engine with an electric motor and kept the shaft.

It took thirty years for factories to be completely reorganized around what electricity actually made possible -- placing power anywhere, rethinking which processes belong next to each other, redesigning the work itself. Only then did productivity increase 2x to 5x.

The delay was not stubbornness. The knowledge required to build the new way simply did not exist yet. It had to be generated through experience.

That analogy framed the second session of the Agent XBE Collective. And it framed a harder question: most organizations using Agent XBE are still running on the shaft. They have replaced the engine -- the manual work, the hours of reconciliation, the reporting grind -- but the structure around it has not changed yet. The question is not whether the structure will change. It is how fast, and who leads.

What the Collective Is

The Agent XBE Collective is not a webinar series. It is not a product feedback forum. It is a working group -- capped, exclusive, and built for the business leaders in heavy materials, logistics, and construction who are shaping how their organizations operate with AI. The concept emerged from an unmistakable appetite among our customers to learn not just from XBE, but from each other.

Last month, we set the foundation. This month, members showed up with live workflows running in production -- and a question about what comes after the obvious wins. What they shared with each other in that room is theirs. What we can share is what it reveals about where this is going.

Replacing the Engine: What April Revealed

The first evidence that the engine is being replaced: in April, XBE customers created over 2,500 Agent XBE sessions.

Two types of work drive that number. Interactive sessions are live, real-time conversations -- a dispatcher sits down with Agent XBE and asks it to build tomorrow's schedule, or an operations lead works through a cost analysis together with the agent. Scheduled sessions are recurring jobs the agent runs on its own -- a daily performance report delivered at 6 a.m., a weekly compliance audit, a nightly reconciliation check -- without anyone prompting it. Think of it as the difference between calling a colleague to ask a question and having that colleague proactively put a briefing on your desk every morning before you arrive.

In April, the split was exactly 50/50. But the trend line underneath is the real story: daily scheduled sessions grew 3x between the first and last week of the month. At that pace, scheduled work could represent 90% of all sessions by summer. The agent is shifting from something people use to something that works alongside them -- running, monitoring, and reporting on its own rhythm.

Interactive sessions, meanwhile, are 40% longer on average than scheduled ones. When people sit down with Agent XBE to think something through, they go deep.

Three categories account for 80% of all work: dispatch planning and execution, monitoring and alerts, and reporting and analysis. Forty percent of sessions resulted in a change to the platform -- not just analysis, but action. And 130 different individuals across the customer base were active in Agent XBE in April.

One number stood out for a different reason: only about 5% of interactive sessions happen on mobile, despite Agent XBE being highly effective on the phone. There is headroom everywhere.

These are impressive numbers. But they describe engine replacement -- faster, better versions of work that was already being done. The factory has not been reorganized yet.

Still on the Shaft: What the Group Saw

When the Collective discussed the factory analogy, the responses were immediate and concrete.

One leader pointed to the time spent planning and scrambling to ensure everyone has the same information -- the old regime of "shut up and do your job." The bottleneck is not the plan itself. It is the informational pockets that never get recorded and therefore cannot be leveraged. The agent can surface and connect that information, but the organization is still structured as if only certain people are supposed to have it.

Another identified two constraints: legacy infrastructure -- core business software still running on a local server -- and the assumption that only management can solve problems. The agent allows end-users to bring prepared solutions rather than problems to leadership, which inverts the traditional information flow. But most organizations have not reorganized around that inversion yet.

A third raised a deeper question: before you can optimize, you have to define what "right" means. Maximizing short-term profit may conflict with doing the right thing for long-term customer relationships. The agent can optimize toward whatever goal you set -- but choosing the goal is a leadership decision the shaft was never designed to surface.

Every organization has a shaft. Most are still building around it.

Signs of Reorganization

If the shaft is the old structure, the show-and-tell was the first evidence of reorganization -- members walking through workflows that do not just automate existing work, but change how work is organized.

We will not detail every workflow that was shared -- that is part of what makes the Collective valuable to the people in the room. But the patterns are worth naming.

Some teams are meeting their organizations exactly where they are -- taking the messy, real-world inputs that dispatchers and planners actually use and letting Agent XBE turn them into finished plans. No migration required. No integration project. Immediate time savings from day one. This is the electric motor replacing the steam engine -- same layout, immediate improvement.

Others have gone further. They have encoded their entire multi-step planning workflows into single commands -- what used to require navigating multiple screens and making dozens of individual decisions now runs as one coordinated sequence. The workflow itself has been reimagined.

The most striking shift: scheduled sessions -- the agent running recurring jobs on daily, weekly, or monthly rhythms without anyone prompting it. Performance scorecards that translate operational gaps into dollar impact. Compliance monitoring that surfaces what is missing before leadership has to ask. Direct, personalized communication to drivers in the field when something goes wrong. The agent is not waiting to be used. It is working alongside the operation on its own rhythm, and the compounding results are measurable.

This is what happens when you start moving the machines away from the shaft.

Whose Judgment Gets Multiplied?

This is the reorganization question. If Agent XBE multiplies judgment, who should get the leverage first?

The value created depends on two variables -- the quality of the person's judgment and the throughput of the work they touch. Prioritize adoption by those who have the best judgment to multiply and the throughput that would yield the biggest impact.

One member sharpened this further with an observation that reshaped the rest of the discussion: intellectual curiosity -- the people who are always asking "what's next" -- is necessarily correlated with good judgment. Judgment is not a single trait. It is the ability to integrate a large number of considerations into a decision. The people who never stop asking questions are the ones whose judgment compounds. They are the ones you want to amplify first.

That framing redefines the adoption challenge. It is not about training everyone equally. It is not about rolling out access and hoping for the best. It is about identifying your best judgment, giving it maximum leverage, and letting the results pull everyone else forward. That is how you reorganize the factory -- not by moving every machine at once, but by placing the most important ones first and letting the new layout prove itself.

What This Means

Three threads emerged from this session that will carry forward.

The leverage on data has changed permanently. When Agent XBE makes it trivially easy to act on data, the return on clean data increases dramatically. Multiple organizations in the Collective have responded by funding data quality as a real workstream -- assigning stewards, forming teams, treating master data hygiene as infrastructure rather than maintenance. Clean data is the wiring of the new factory. You cannot place machines in new configurations if the power lines are unreliable.

The question is no longer whether the agent works. It is whose judgment should power it. The Collective is producing clear evidence that the right person with agent leverage creates disproportionate value. Identifying those people -- and giving them maximum leverage first -- is now the strategic challenge.

The factory is still running on the shaft. Most organizations are doing the equivalent of replacing the steam engine with an electric motor and keeping everything else the same. The Collective exists to accelerate the thirty-year learning curve -- to generate the knowledge required to reorganize, together, instead of waiting for each company to discover it alone. What the members are sharing with each other in these sessions is how to start.

The Agent XBE Collective meets monthly. Membership is reserved for the business leaders in our XBE community who are driving AI adoption in their organizations. To learn more about Agent XBE and what it can do for your operation, reach out to your XBE account team.