Manufacturers face a dizzying array of potential problems around the machines they produce, and it’s challenging to track down issues. This isn’t just nice to know. It’s crucial information, often tracked manually today by human auditors in spreadsheets. In some cases, failing to understand when there is a faulty part could result in costly recalls, and in the most extreme cases, deaths and lawsuits.
Enter Axion Ray, an early stage startup that is using machine learning to track these issues in unstructured data to build a picture of potential problems before they get out of hand. Today the company announced a healthy $7.5 million seed round.
“What we’ve done is build a new artificial intelligence platform that helps manufacturers get ahead of their major risks like recalls by tapping in and synthesizing unstructured data in new ways that up until this point haven’t really been touched,” Axion Ray co-founder and CEO Daniel First told TechCrunch.
He says that the unstructured data comes from human users and separates his company from those that have come before him.
“In traditional machine learning, and many of the companies that have come before us, much of the focus in manufacturing has been in highly structured datasets like installing cameras on the manufacturing line, or looking at sensor data to predict an engine failure.”
“But what’s exciting about Axion is that we can leverage huge amounts of unstructured data, things like [chatter] coming from service or dealership networks, where most of the data is technician observations, and is found in comments and issues and troubleshooting data coming from humans.”
First worked as a McKinsey consultant for several years before launching the company, and saw first hand how manufacturers were struggling to recognize potential problems before they really blew up on them. He also observed that technicians working on these machines were seeing problems months before the companies realized there was a broader issue, and the idea for Axion Ray began to take shape.
“It became obvious that there was a huge opportunity to enable the detection and the flagging of the earliest warning indicators, and that could help people detect that there are risks months earlier.”
The company was founded in 2021 and already is working with customers like Boeing, Penn Engineering and Cummins. First didn’t want to share the number of customers just yet, but it’s clear some big players are interested in what his company is doing.
With almost 20 employees, the startup is hiring, especially engineers and employees with a specialty in machine learning. First says building a diverse workforce has been a priority from the start.
“Even though we’re a small team, we have dedicated full time colleagues who are responsible for ensuring we’re building diverse candidate pipelines and hiring practices from day one. We were also thrilled that we were able to partner with Inspired Capital as our co-lead investor, which is one of the largest female-led venture funds in the country,” he said.
Today’s $7.5 million investment was co-led by Inspired and Amplo along with Boeing, Tinicum Venture Partners and industry angels.