Making the Complex, Simple: Sunny Han on ERP, AI, and the Human Side of Manufacturing
At FABTECH 2025, David Lechleitner of LevelUp Consulting invited Fulcrum CEO Sunny Han to join as a guest speaker during his talk: “Making the Complex, Simple: The Impact of ERP UI, UX, and AI on Your Employees' Work Satisfaction.” The session quickly turned into an open Q&A that revealed how Fulcrum is rethinking the role of ERP in manufacturing.
The questions circled around a few major themes that manufacturers everywhere are wrestling with: where to invest in AI, how to design software people actually want to use, and how to bridge generational differences on the shop floor. Sunny’s answers reflected Fulcrum’s philosophy: focus on what truly drives throughput, even if it isn’t flashy.
Listen to the excerpt:
Throughput as the North Star
Fulcrum’s guiding principle is simple: help manufacturers make more with the same people and machines. As Sunny put it, “your machines are the mechanical muscles of your shop, and Fulcrum is the mechanical mind.”
That mindset drives how Fulcrum decides where to invest. The priority isn’t on eye-catching features that look good in a demo but don’t change much on the floor. Instead, the focus is on deceptively simple tools that eliminate bottlenecks.
For example, Fulcrum is working on tools that automatically scan emails for order details and feed them into the system, making information instantly available to the rest of the shop. The same principle applies to document automation in development — first article inspections, certificates of conformance, and reports can all be generated instantly from customer templates.
“These things don’t sound super cool,” Sunny noted, “but they free people from the low-value work that eats up hours every week. That’s what drives throughput.”
On average, shops running Fulcrum see a 35-40% increase in throughput after implementation — same machines, same people, just better coordination and less time spent on paperwork, double-checking, or chasing information across multiple systems.
Design That Earns Engagement
ERP adoption lives or dies by user experience. Sunny shared how his early career implementing legacy systems revealed a troubling trend: as interfaces became harder to use, operators entered less data, and ERP systems wasted away as silos. Fulcrum’s counter-approach has been to invest 4x more in design than the industry norm, ensuring operators gain immediate value each time they enter data.
One example: when an operator enters a checkpoint in Fulcrum’s Job Tracker, the system immediately flags if it’s out of tolerance and (soon) will highlight it on the drawing. Instead of a note disappearing into a stack of paper, the operator gets instant feedback. Interactions like these build trust and keep data flowing into the system instead of spreadsheets or side notes. That usability keeps data density high, enabling advanced AI, automation, and reporting.
Bridging Generations on the Shop Floor
The Q&A also highlighted the challenge of serving a multigenerational workforce. Sunny described it as a kind of social contract. For veteran machinists, Fulcrum promises to capture their knowledge and reduce paperwork so they can focus on their craft. For younger employees, the priority is intuitive, real-time tools that feel familiar — simple search bars, mobile-friendly screens, and live updates.
The balance, Sunny noted, is less about age and more about shop culture: successful adoption happens where companies see employees as assets to be empowered, not just resources to be managed.
Unlocking AI in Manufacturing
While Fulcrum processes billions in orders annually, Sunny pointed out that the industry is still scratching the surface: “We estimate we only have about 8% of the necessary data to do the really cool stuff.”
Data Density
Each year, an estimated 1.9 billion drawings are created, revised, and reproduced. Yet only 0.000001% of them are public and available to train AI models — nowhere near enough to build reliable systems. Without access to real data, models hallucinate. And training on drawings alone doesn’t work. Shops uniquely know how to make parts, whether that means a job goes flawlessly or ends up as scrap. Both outcomes are essential for training useful AI.
Even more knowledge is buried in spreadsheets, Word docs, and years of email threads between customers and engineers. As Sunny noted, simple small details — like how a bend line gets discussed with a customer — often carry decades of hard-earned tribal knowledge. This type of unstructured data is not only missed by ERP systems, but is a complete blind spot for today’s models relying on publicly available data.
“As we increase that density,” Sunny explained, “we can do cool AI stuff, we can do cool reporting stuff, we can do cool automation.”
The Danger of Customization
Most ERP systems compound the problem. Heavy customization leaves datasets inconsistent and scattered across custom fields and siloed modules. Adoption is often so low that critical information dies on paper or lives in isolated spreadsheets, inaccessible to any model.
Customization may let a manufacturer mold ERP to its process, but the trade-off is an island of data with no harmony — no patterns of actions, reactions, iterations, or decisions for AI to learn from. For AI to work in manufacturing, data structure harmony is essential.
What Comes Next
The next leap requires anonymized proprietary data — CAD models, work instructions, production results — combined with AI to deliver breakthroughs in scheduling, supply chain, and design efficiency.
That’s why Fulcrum is focused on making data entry easy, consistent, and cleanly structured today. It’s the foundation for what will become Fulcrum Intelligence tomorrow.
Transcript
Sunny:
We’ve hired a bunch of AI experts, including engineers who used to work at Microsoft and OpenAI. They’re helping us fine-tune models, and we’ve already invested a few million dollars experimenting with different approaches. Some of it has turned into genuinely useful tools. I’m happy to answer questions about what we’ve learned. It might not all be relevant, but hopefully some of it is helpful.
David:
Great. I actually have a couple questions for you. Every company has to make choices about where to invest dollars, but software companies probably face that challenge more than most. How do you determine where to invest, and where does AI fall in that hierarchy of needs?
Sunny:
Internally, we think of laser cutters, waterjets, press brakes, and CNC machines as the mechanical muscles of our customers’ businesses. And we think of Fulcrum as the mechanical mind. Whatever your machines do to automate physical work with more precision and fewer labor hours, we try to do the same for coordination, communication, and decision-making.
That’s our north star. We build software that increases throughput. We track how much more our customers produce after implementing Fulcrum, with the same people and same machines. Whether that metric is perfect or not, we deeply believe it’s the best measure of how we serve our customers.
Right now, the average customer sees about a 35 to 40 percent increase in output simply from coordinating better. And we think we can push that further.
There are a lot of AI features that look flashy but don’t actually move throughput. And then there are features that sound boring but matter enormously. For example, we’re working on something that automatically reads all your emails and pulls relevant information straight into the system. Not flashy, but it makes information available faster for everyone else in the shop.
We’re also using AI for document generation, first article inspection packets, auto-reports, certificates of conformance. You drop in a customer template, and Fulcrum automatically searches the system and builds the document. Again, it doesn’t sound exciting, but it automates work that consumes time without adding value.
Most of the features that truly increase throughput aren’t sexy, but they have a huge impact. That’s why throughput is our north star. Can you make more stuff without spending time on paperwork, data entry, double-checking, or digging through information? Can we connect employees directly and give them the right information instantly?
We’re working on vision-system tools where you record how a job should be done, and we index that automatically. Or relating geometric models with 2D prints and bubbled drawings so quality checks happen in one place instead of three separate documents.
Again, none of this is glamorous, but everything is prioritized based on one question: Will it help you produce more with the same fixed resources?
David:
Thank you. One last question, drilling back a bit. In my research, one thing that kept coming up was the importance of UI and UX, and the direct correlation to job satisfaction. I think Fulcrum is at the forefront of UI and UX design. I’m ERP-agnostic, but I’ll say you all do a great job there. What was your progression to get there? There’s been a lot of bad UI and UX in this industry.
Sunny:
Early in my career, my first job was implementing systems like JobBOSS, E2, Epicor, Acumatica. I didn’t know anything about manufacturing; I was just a kid who could write SQL queries. Over ten years, I saw a pattern emerge: the amount of data people put into their ERP systems dropped every year.
At first, everything went into the ERP. Then scheduling got a little harder, so scheduling moved to a spreadsheet. Estimating got more complex, so estimating moved to a spreadsheet. Quality schedules moved to spreadsheets. Slowly, the ERP that was supposed to be the central source of truth just atrophied. It held less and less of the real operational data. Everything became disconnected.
Then you’d walk into a production meeting and no one really knew what was going on. The whole point of having centralized information had evaporated. The only way these systems still functioned was by being draconian: “You have to enter this because we said so,” not because the system gave anything back.
When we launched Fulcrum, we asked: Is this erosion inevitable? And if not, how do we reverse it?
The answer was design.
If we make the system pleasant, intuitive, and genuinely helpful, people will want to use it. If entering a checkpoint immediately highlights out-of-tolerance dimensions on the drawing, the operator gets something out of it. That beats writing numbers on paper, handing it off, someone punching it in days later, and three weeks after that someone says “you messed up.” That workflow is terrible.
So the core idea became: make every user love using the system. If the experience is good, data density increases again. And when data density increases, we can do the cool AI stuff, the cool reporting stuff, the cool automation stuff. But without good design, you can’t do any of it.
We spend four times more per dollar of revenue on design than other companies. We get laughed at for it, but we believe in it.
David:
I love that. Last question. We’re seeing four or five generations all working with the same technology. New employees grew up with iPhones and Apple Watches. People like me grew up with blue-screen terminals. How do you address that gap?
Sunny:
Each group wants different things. More experienced workers know a ton, and many of them want to pass that knowledge on as they plan for retirement or transitions. So our “social contract” with them is: we’ll capture your knowledge and reduce your paperwork so you can focus on your craft.
Younger workers expect everything to be live, intuitive, and searchable. They don’t want outdated information. They don’t want to look unprepared. They’re used to typing anything into a search bar and instantly getting results.
So for them, the social contract is different: we’ll make the system intuitive, fast, and aligned with modern software patterns.
Balancing those expectations is easier than you’d expect. The bigger challenge is cultural. If a shop doesn’t care about its employees and treats them as interchangeable, adoption is brutal. If a shop values its people and sees them as assets, implementations are fast, smooth, and incredibly successful.
TL;DR: Fulcrum’s mission isn’t about chasing trends. It’s about building a system that actually helps people on the shop floor work smarter, faster, and more connected. As Sunny put it, “Our North Star is throughput — can you make more stuff with the same fixed resources? That’s how we win, and that’s how our customers win.”
