How This Manufacturer Used Data to Increase Efficiency and Cut Costs This is what happens when machine learning comes to the factory floor.

Inteva Products, a multibillion-dollar global manufacturer of auto parts, wanted to improve its products and sniff out manufacturing inefficiencies. So it ran an experiment: It hired Sight Machine, a 4-year-old San Francisco-based company that turns complex workplaces into crunchable data. The system works like this: For any part of a factory that Sight Machine wants to monitor, it sets up cameras and sensors and finds ways to pull in live feeds from any internet-connected devices, right down to a shop's HVAC system. Then the startup's software takes over -- pairing the data it's picking up from the cameras (say, how quickly employees are moving) with everything else it's learning -- to find out when and why production is lagging. "People making cars, shoes, drugs -- they all want to know what's wrong with their operation, and our software can tell them," says Sight Machine's CEO, Jon Sobel, a veteran of Tesla and Yahoo!

Inteva did a trial run to see if Sight Machine could identify ways to reduce the amount of scrap coming off the production line. It was a success, says Inteva CIO Dennis Hodges. He won't say how much he saved, but consider this: Sight Machine costs $50,000 to $100,000 to set up, plus a monthly subscription fee. To justify the cost, the savings have to be real. Now Inteva is expanding its use to see if the startup can help improve Inteva's injection-molding process. The data awaits.