The challenge
The test waits for the expert
Testing an electric drive means recording voltages, currents, torque and speed for hours or days, and every result depends on the setup behind it. A mistake in the wiring or the settings can leave a recording that looks normal while the results are wrong.
When a measurement looks wrong, the technician at the test stand calls an engineer. The engineer may not be on site, so the test waits while the problem travels through phone calls and sent files. The technician cannot relay everything the engineer needs, and the engineer cannot see everything the technician sees. Running tests safely and efficiently is the technician's job; troubleshooting the data is not.
A diagnosis that takes minutes at the stand takes 30 minutes to two hours remotely. Sometimes the problem is never fixed, and the test, or part of it, is lost.
Asked how long a signal or wiring problem takes to diagnose without the app, Mitchell said it depends on who is looking, and whether they know what problem they are looking for. For an engineer at the test stand who does, it is a matter of minutes. For everyone else:
A lot of people might not even know what to look for, and this is where HBK's knowledge being put in an app helps you know what to look for every time.
The solution
The expert's knowledge, in one place, deployed and working for everyone
moveOS, MOVEdot's operating system for engineering, connects a team's test data, analysis methods, tools and knowledge so that AI agents can work across all of them. At HBK, Mitchell used it to put his troubleshooting method, and the equipment knowledge behind it, into the platform. The agents build and run the analysis; he defined each check and decides what the app reports.
The expert's knowledge lives in the platform. The checks Mitchell runs from experience are saved as skills and workflows, so they are applied to each new recording automatically. That knowledge is specific to HBK's tools and equipment: the app carries the supplier's expertise, not only one engineer's, and finds faults that users might not know to look for.
These are things that I, as an expert, have a gut feeling for diagnosing. This [the moveOS app] has put logic and science to it.
It finds what nobody was looking for. Mitchell troubleshoots a lot of data, and some of these faults he would not have found on his own.
I would not have found some of the things that I found with this tool. I would have suspicions, but if my tech is running me these tests and sending me results, I'm not looking for them.
Problems noted in other tests turned up in tests he would not have suspected, before they became issues.
The diagnosis goes to the technician. When a recording is uploaded, the checks run on their own. Inside moveOS, Mitchell built an app to tell the technician what is wrong and what to change, without needing to call an engineer for the common cases. When a flagged result needs an engineer, the app gives the technician the vocabulary to describe the problem, and the engineer can have the agent further analyze the data and give more details on the result, so the judgment call is made quickly.
Everything is connected. The same app is a centralized place where the engineer follows every test, and the platform can send an overview report by email after each one, whether the checks pass or fail, and why.
Three checks from HBK's recordings
Each one is slow or hard to find by eye.
Data quality: had each test point settled?
The test points had not fully settled, so the results at those points could not be trusted.
Data quality: had each test point settled?
The test points had not fully settled, so the results at those points could not be trusted.
Visually, I didn't even pick this up, and then I dug into it and I was like, ‘Oh yeah, no, the test points were wrong.’
What it led to. Mitchell went back into the data and confirmed the finding. He had “been looking at that data for years and never noticed it.”
Wiring: are the phases connected in the right order?
The three phases were measured correctly but wired in the wrong order. The app showed which signals to swap.
Wiring: are the phases connected in the right order?
The three phases were measured correctly but wired in the wrong order. The app showed which signals to swap.
Each channel carries a valid signal, so nothing looks wrong on its own; the error only shows in the combined results. An expert might catch it at a glance; someone running tests back to back will not. The correction is stated in terms a technician can act on at the test stand.
Test rig: is the rig itself affecting the result?
A mechanical resonance in the rig near 400 Hz, visible in the torque and speed signals.
Test rig: is the rig itself affecting the result?
A mechanical resonance in the rig near 400 Hz, visible in the torque and speed signals.
The resonance comes from the rig, not from the machine under test. The app recommended avoiding the affected operating points in the next test.
The impact
The fault is caught before the test is lost
A diagnosis that took a remote engineer up to two hours now takes about 5 minutes, up to 24x faster. moveOS runs up to 21 expert checks on a recording, and the answer reaches the technician at the test stand.
The speed matters because of what is at stake: an electric-drive test ties up a dynamometer, an engineer and a technician for hours or days, and a setup fault found afterward means running it again.
- The fault is caught before the long run. A short sweep is checked first, so wiring and settings are confirmed before a two-hour or three-day test starts, and fewer points need retesting. A couple of hours of lost test-stand time a week adds up to around $50K a year.
- The technician can act without waiting. Each finding comes with the fix: which signals to swap or which setting to change.
- HBK's expertise reaches every test stand. Years of troubleshooting experience with HBK's equipment are applied to every recording, wherever the engineer is.
- It finds what nobody was looking for. One check flagged a settling problem in data Mitchell had been looking at for years; others surfaced faults in tests he would not have suspected.
- The expert's time goes where it is needed. The engineer steps in for the cases that need judgment.
moveOS is what made this possible: it turned years of Mitchell's troubleshooting experience into an app and workflows that can be deployed, shared and used across the organization.
What comes next
Measurement and AI, built together
HBK's equipment captures the measurement, and HBK's knowledge of that equipment is what the checks are built on. moveOS turns both into a diagnosis the whole test floor can use. As partners, HBK and MOVEdot bring the two together: HBK's acquisition data and expertise, on an AI platform built for engineering.
Mitchell built all of this in two weeks, with support from MOVEdot's engineers. If two weeks produced this, imagine what HBK, its customers and engineering teams in general can build with moveOS as the foundation of their engineering work. The next stage, in Mitchell's words, is taking the same approach from troubleshooting to analysis.
Put AI to work across vehicle development
moveOS gives AI agents the engineering tools, data and knowledge to do the work.



