Manufacturing Connected Interview with Sunny Han: AI & Shop Floor Applications
Published on Manufacturing Connected | Read the full article →
Shop owners have questions about AI: Do you need an IT team? How do you adopt AI without violating CMMC or ITAR? What happens to tribal knowledge when your best guy retires?
Manufacturing Connected's Editor-in-Chief Stephanie Hendrixson asked Fulcrum Founder & CEO Sunny Han to answer some of the common questions she is seeing come out of conversations with machine shops and fabricators lately. Here's a quick look at where Sunny lands on each one.
You don't need an IT department. The machinist who learned CNC without a computer science degree can learn this too. What you do need: curiosity, and a willingness to be uncomfortable for about 90 days — the tools themselves are not too hard to pick up, but they will rapidly share new information that will take some time to process and respond to. AI can quickly reveal where the time actually goes and which jobs actually make money. Sometimes it can end up revealing that 20% your instincts were wrong about.
CMMC and ITAR require careful consideration and planning. AI that handles controlled data needs to run inside a walled garden. Azure Government, AWS GovCloud, self-hosted on your own network. The problem is that standing up that infrastructure alone is a seven-figure project. Sunny's argument: the same reason you don't build your own shop management software from scratch is the reason shared infrastructure makes sense here. What's impossible at the single-shop level is affordable at the network level.
Quoting with incomplete information is a solvable problem. When a customer walks in with a worn part and no clean model, the AI doesn't get stuck. It pulls your production history, your pricing decisions, your shop floor data, and it learns every time you quote. The goal is more than just a clean CAD file. It's your institutional knowledge, connected in real time.
Tribal knowledge already exists outside people's heads. It's in your job histories, your scrap rates, your quoting patterns. The knowledge is locked in your best people's brains. No human can hold all of it at once and see the patterns, but oddly enough AI can. This is why it's so beneficial to give AI access to watch and learn as your shop runs.
AI is software that learns you. Every piece of software you've ever used, you had to adapt to. AI flips that, so you ought to consider treating it like a new hire, not a new tool. You wouldn't expect a new employee to run your best job on day one. After 90 days they're useful. After a year they're valuable. After five years they're training the next hire. Same principle.
Transcript
The full article is worth reading, especially if you're fielding these questions from your own leadership team right now:
