Five discrete manufacturers. Five real problems. Five working solutions built on live production data in 48 hours.
Every manufacturer we talk to has the same frustration buried somewhere in the conversation: their software almost fits. The MES handles work orders, scheduling, and OEE. But the workflows that are specific to their operation, the ones that live on whiteboards and clipboards and tribal knowledge, those never quite make it into the system.
So they build workarounds. They create spreadsheets. They page people over intercoms. They stand in front of paper charts every morning and manually calculate whether they're having a good day or a bad one.
In March 2026, we decided to see what would happen if we gave five of our customers two days, a team of engineers, and AI-assisted development tools, and asked them to close those gaps themselves.
The results were better than we expected.
Production Lab is a two-day, hands-on building event where MachineMetrics customers work side by side with our engineering team to design and build custom applications on the MachineMetrics platform.
This year's event took place on March 11 and 12 near our headquarters in Northampton, Massachusetts, during a week-long engineering onsite. Five customer teams attended, each arriving with a specific operational problem and one shared goal: walk away with something they could actually use on the shop floor.
The tools they used: Max AI for data exploration and prompt development, Lovable and Cursor for AI-assisted application building, and Carbide, MachineMetrics' custom application builder, for deploying production-grade applications directly into the platform.
Every team shipped a working solution tied to real, live machine data by the end of day two.
Harvey Performance makes industrial cutting tools for aerospace, medical, and electronics customers. Their shop runs complex, made-to-stock production with hundreds of job families, and a persistent problem: when a job ends, machines sit idle until someone physically arrives to start the next setup.
Harvey had the technology to do automated changeovers between jobs sharing the same tooling. What they lacked was a system to plan and orchestrate it. Without it, weekend utilization would drop significantly by Sunday night and take days to recover.
In two days, they built the Stacker Tracker: a live supervisor dashboard that shows every machine's active job, all queued jobs, and projected runtime through the lights-out window. The system calculates dynamic setup times based on Harvey's proprietary batch code structure, flags labor conflicts, and lets supervisors assign operators to setups with specific start times. If the setup doesn't happen, the schedule auto-reverts. No manual cleanup required.
The business case: a seven-figure cost savings initiative and multiple millions in additional revenue from recovered capacity.
"Within a day and a half, we were able to design something that would probably take months to have created previously, and it's tailored specifically to our operation." — George Burleson, Director of Manufacturing Analytics, Harvey Performance
Zygo makes ultra-precision optics for semiconductor, defense, and laser fusion applications. Every day, their management team stands in front of whiteboards, paper charts, and printed graphs to determine whether their operation is on track.
Their goal: make all 15 key metrics across Safety, People, Quality, Delivery, and Value readable in 10 seconds from 10 feet, at every level of the organization.
They built a fully digital, multi-tier management system in Lovable. It drills from the enterprise level all the way down to individual machine cells in a few clicks. Max AI generates summaries and flags patterns at each tier. A configurable ticketing system brings safety concerns, maintenance requests, and continuous improvement ideas into one place, so nothing gets lost across disparate systems.
It touches every person in operations on day one.
"Things like the digital tier board system we built here in this two-day event wasn't even on my radar for things that MachineMetrics was able to achieve. With the advancement in AI and Carbide, MachineMetrics is bringing new avenues, new value, and flexibility every single day." — JD Smith, Director of Operations for Optics, Zygo/AMETEK
Pindel runs a government contract on four CNC machines, 24/7. Different operators were adjusting tool offsets at wildly different rates, but no one could see it. The machines were connected. The data was flowing. It just wasn’t being captured anywhere meaningful.
Thomas Deslongchamps, Director of Training and Continuous Improvement, decided to change that. Using Lovable and the MachineMetrics Carbide Application Builder, he built a custom tool offset analytics application almost entirely on his own. It visualizes offset adjustments per tool over time, correlated to part count and color-coded by operator. For the first time, Pindel could see which operators were outliers, when adjustments were clustering, and what patterns were emerging across machines.
"We're looking to move away from the more preferential operational status quo to a more data-based status quo. Could we make fewer offsets and still stay within the quality parameters?" — Thomas Deslongchamps, Director of Training and Continuous Improvement, Pindel Global Precision
A MachineMetrics engineer working alongside Thomas used a custom AI agent to scan the underlying code and found a more efficient data access path, one that unlocked exactly the functionality Thomas needed and opens the door to even broader use cases ahead.
The capability had always existed inside the platform. It took the right customer, the right use case, and the right toolchain to bring it to life. What Pindel built isn’t just solving their problem. It’s defining what’s possible for every precision manufacturer, asking the same invisible question.
Flexco manufactures conveyor solutions. At their Downers Grove facility, forklifts (called "jeeps" internally) are critical assets with zero visibility. Material requests go out over the intercom. Drivers have to remember what was asked and where. In 2025, that blind spot cost an estimated $225,000 in lost production time.
Jaimeson Aufderheide, a Continuous Improvement Specialist with no prior software development experience, built the core application himself using Lovable. By the end of day two, Flexco had a fully embedded dispatch system complete with an operator request interface, a live driver queue with zone-based prioritization, a manager dashboard showing real-time and historical ticket performance, and a configurable widget that rolled it all into a single deployable tile.
The Lovable and Cursor codebases were merged through GitHub into a unified data model. That cross-tool collaboration pattern emerged organically during the event and was not something anyone had planned for.
"Two days ago, if I asked you, 'you got it?' what would you have said?" "I don't even know where to start." — Exchange during the Flexco showcase
Johnstech makes semiconductor testing equipment. Their problem: operators were either replacing tools too early, changing everything at shift start regardless of wear, or running them too long and producing bad parts.
The solution was a tool life dashboard that shows all tools ranked by actual engagement time and color-coded for attention. But the real surprise came mid-build: the team discovered that Johnstech already had a tool change sub-program running on every machine since day one of their MachineMetrics deployment. All of that historical data was already in the system, just unprocessed.
They backfilled 90 days of tool life history retroactively. No new hardware. No new configuration. Just processing what was already there.
The second project closed a gap that had existed for 10 to 15 years: a live ERP integration with Microsoft Dynamics 365 that compares planned versus actual production run time for every operation, every day.
"That's been something we've just never known for 10, 15 years. And now we have this information." — Dan Sheehan, IT Manager, Johnstech International
MachineMetrics is not just a tool that manufacturers look at. It's a platform they build on.
That's not a positioning claim. It's what happened across five companies, five different operational problems, and five different skill levels over two days. A CI specialist with no coding background shipped a production-embedded application. A customer built the first commercial tool offset analytics app, on the platform, almost entirely on his own. A 10-to-15-year gap in production intelligence was closed with data that had been sitting in the system the whole time.
The execution gap is real. It lives between what your ERP plans and what your shop floor actually does, and it costs manufacturers margin, capacity, and hours every single shift. Closing it requires software that fits how your operation actually runs, not a generic template you adapt yourself.
"It's been great to work shoulder to shoulder with the engineers, combining their software development experience with the operations side to see solutions come to life in real time. And then being able to walk two feet to the left and have a conversation with one of the executives about the strategy of the company and what technology is coming down the road. The flexibility that drew me to MachineMetrics is on full display here at this event." — JD Smith, Director of Operations for Optics, Zygo/AMETEK
That's what Production Lab is for. And we're already planning the next one.
If you're a MachineMetrics customer with a workflow that doesn't quite fit the standard toolset, we want to hear about it. And if you're a manufacturer still running that process on whiteboards and clipboards, let's talk about what two days could do.