Telematics has been standard on new construction equipment for years. Most contractors can see where their machines are and how many hours they've logged. The harder question is what to do with that information, especially for fleets that mix equipment from several manufacturers, each with its own portal, data format and definition of something as basic as an engine hour.

Equipment makers, rental companies and technology firms are now pitching artificial intelligence as the answer: software that sifts through machine data and flags what needs attention, rather than leaving fleet managers to dig through dashboards.

Trackunit is one of those companies. Its platform connects more than 6.5 million off-highway assets, according to the company. Roads & Bridges spoke over email with Federico Rio, Trackunit's senior vice president of product and pricing and a former Caterpillar engineer, about what contractors can realistically do with connected-machine data, which numbers matter and which are noise and how fleet management may change over the next decade.

This interview has been edited for length and clarity.

R&B: Connectivity is becoming increasingly common across construction equipment. At this point, what are the most meaningful things contractors can actually do with connected-machine data that they couldn't do before?

FR: The change I notice the most, when it comes to acting on machine data, is small, but it tells you a lot. It used to be that when a machine had a problem, the first move was to send a truck. Now it's to open the screen and look before anyone drives out.

That same shift plays out across a whole fleet. You catch a machine idling before it burns through your fuel budget, send a rental back before you've paid for it sitting still, and fix a fault between phases instead of when it quits on you mid-shift.

And the encouraging part is that we're finally seeing this data get put to real use. There's always a curve with new technology, and the early adopters are already pulling ahead.

What lets them do this is AI. A connected machine throws off more data than any analyst could ever keep up with, but you can now put AI agents on top of it, software that reads the data for you, and a single one can do the work of 10 analysts. The companies doing this are getting real results.

R&B: You argue that the real value comes from connecting an entire fleet rather than individual machines. What are the biggest challenges contractors face when trying to connect and manage a mixed fleet with equipment from different manufacturers?

FR: I came up as a mechanical engineer, and even inside a single company the data rarely lined up. Now picture a contractor fleet: maybe 1,000 machines from ten manufacturers, each with its own portal, its own login, its own definition of what a fault or an hour even is.

But the real challenge isn't collecting all that, and it isn't the technology. It's that the sheer scale makes the analysis fragile.

You can write a rule to watch the data stream, but the moment something looks slightly off, the rule breaks. And to cover every situation you'd need a thousand rules for a thousand exceptions, which nobody can maintain.

This is where AI genuinely changes things. It does the analysis for you, it bends to the exceptions instead of breaking on them, and it even catches the ones a hand-written rule would miss.

What you're left with is a single view you can actually trust, where you're reading your own numbers instead of translating someone else's.

R&B: Connected equipment can generate enormous amounts of data. What information is genuinely useful to a fleet manager or equipment owner, and what data tends to create more noise than value?

FR: Honestly, you need far fewer exotic data than people assume. Just hours, location, fuel and fault codes will take you a long way, and I'd frame it around three pillars: equipment management, safety and productivity.

Equipment management is really your financial KPIs. You can depreciate the fleet in real time based on hours, which is gold if you're a CFO, or reconcile fuel bills against actual usage and see straight away where you're losing fuel.

Productivity comes down to utilization, and it tells you about your processes, not your people. If machines are sitting idle, it's usually not lazy operators, it's four machines waiting on a cement truck that never showed up.

Safety, on its own, is worth it for access management alone.

The noise is everything that doesn't point back to one of those three. If a number doesn't change a decision you'd make, you don't need to be staring at it.

R&B: As construction equipment becomes more connected, what do you think will change most about the way contractors manage their fleets over the next five to 10 years, and what aspects of fleet management do you think will remain largely unchanged?

FR: The biggest change will be AI finally letting us do the things that were always worth doing but that you could never realistically get done with people.

Think of it as an army of virtual employees: AI agents that quietly work through your data and hand you the insights, again across those same three pillars of equipment management, productivity, and safety. Instead of doing the analysis himself, the equipment manager gets it done through them.

It changes how you work in the field, too. Rather than tapping through screens, you'll simply talk to the whole ecosystem of data, telematics and everything around it, by voice on your phone.

What won't change is that this is still a physical business. Someone still has to run the machine and make the call on the ground. All of this just makes those people faster and better informed.

R&B: The ISO 15143-3 standard, developed with AEMP, was meant to solve the mixed-fleet data problem. Why hasn't it been enough on its own, and where does it fall short for contractors?

FR: The standard was a step in the right direction, but on its own it hasn't solved the problem, for a few reasons. One: It isn't prescriptive enough. Too many data points are left open to interpretation. Fuel is a good example: one OEM reports "liters in the tank," another "% full," another "% empty." All of it is technically compliant, which makes the data very hard to harmonize.

Two: It doesn't harmonize depth, breadth, or latency. The recommended categories are enough for basic fleet management but lack the depth you need to go any further, into advanced condition monitoring, productivity, and safety. And because the standard says little about frequency, one provider can send a data point every 15 minutes and another once a day, and both still call it "compliant." A standard that allows that isn't really supporting the end user's needs.

Solving those problems takes a data source with intimate knowledge of the machine: telematics developed together with the OEM, and a data transfer system that gives you the full depth of the data and the semantics needed to interpret it.