He spent years watching enterprise AI pilots die on a slide. Then he built the company that ships the part everyone skips - the fix.
A surgical device was still hurting patients. The manufacturer had already declared the problem solved, closed the ticket, moved on. Axion's software read the field reports the humans had skimmed and surfaced two distinct root causes nobody had connected. That is the entire company in one anecdote - the gap between "we fixed it" and "it is actually fixed."
Daniel First runs Axion (you may know it as Axion Ray), an AI quality intelligence platform that sits like an air-traffic-control tower over a manufacturer's product line. It ingests the messy stuff - technician notes, IoT signals, warranty claims, customer complaints - and flags the engineering issue that is quietly metastasizing across a fleet of cars, jet engines, or medical devices. The pitch is blunt: every day a manufacturer stays unaware of a defect can cost millions, and a full recall can cost a great deal more.
What makes the story interesting is not the model. It is the diagnosis behind it. First spent his McKinsey years inside enterprise AI projects that looked dazzling in the prototype and then quietly expired. He concluded the failures were rarely about math. They were about three things: missing domain expertise, processes the AI did not fit into, and no clear path from a clever output to an actual decision. Axion is his answer to all three, built on purpose to close those gaps rather than wow a boardroom.
So Axion hires the unlikely staff for an AI startup: aerospace engineers, medical-device specialists, people who have signed off on parts that fly and parts that cut. They map how a real quality team works, then teach the software to automate the repetitive analysis step by step. Human-in-the-loop is not a slogan here. It is the architecture. The engineer stays in the chair; the AI does the reading no human has time for.
The customer list reads like an industrial hall of fame - Boeing, Cummins, Baxter International, Newell Brands, DENSO, and Pratt & Whitney. RTX, the parent company behind Pratt & Whitney and Raytheon, liked the idea enough to do something rare: invest in Axion through RTX Ventures and use the product. Investor and customer at once. Bessemer Venture Partners led the $17.5M Series A in March 2024, pushing total funding to $25M.
First's framing of the whole field is refreshingly un-grandiose. He does not talk about AGI or replacing engineers. He talks about learning speed. In his telling, the next era of American manufacturing belongs to whoever can hear what customers are experiencing and reshape the product fastest in response. Quality, the least glamorous word in the building, becomes the competitive edge.
It helps that he has lived on both sides of the lectern. Before Axion he taught Enterprise AI Strategy at Columbia Business School - explaining to executives how to actually deploy this stuff - and led AI engagements in automotive and aerospace at McKinsey and its AI arm, QuantumBlack. The academic credentials stack high: a bachelor's from Yale, a master's from Cambridge, and a master's in data science from Columbia Engineering. The career arc is the rare one that runs from theory straight into the factory floor.
There is a tidy irony in the name. A founder named First, building first-warning systems for the world's most consequential products. The earliest signal, caught earliest, by a guy whose surname is the word for it.
Most AI founders chase the demo. The flashy generation, the eerie chat, the moment the room gasps. First chased the opposite - the unglamorous middle of a workflow where a real decision either gets made or does not. His core observation from the McKinsey years is almost anti-climactic: the models were already good enough. What kept enterprise AI from sticking was everything around the model. Nobody on the team knew the domain deeply enough to trust the output. The process had no slot to drop the insight into. And there was no clean line from "the system noticed something" to "a person did something about it."
That diagnosis is the spine of Axion. Instead of selling a smarter brain, First sells a closed loop. The platform reads the unstructured exhaust of a manufacturing operation - the service tickets written in shorthand, the sensor streams nobody has time to chart, the warranty claims filed in a dozen inconsistent formats - and it does the synthesis a human team would do if a human team had a thousand hours a week. Then it routes the finding to the engineer who can act, in the language that engineer already speaks.
The economics are stark enough to make the pitch easy. Poor quality, by First's framing, drains trillions from manufacturers globally every year. A recall is the nightmare ending, but the slow bleed is worse - the weeks and months a defect spreads quietly through a product line before anyone connects the dots. Each of those days carries a price. Catch the pattern early and the math flips from catastrophe to footnote. Customers have reported an average 27% cut in downtime and a 16% reduction in warranty and service costs, the kind of numbers that survive a CFO's skepticism.
The customer roster does a lot of persuading on its own. When Boeing, Cummins, Baxter International, Newell Brands, DENSO, and Pratt & Whitney are all in the same sentence, the common thread is consequence - aircraft, engines, medical devices, the products where a missed defect is not an inconvenience but a headline. These are also the hardest buyers in the world, institutions with their own armies of quality engineers and decades of process. They do not adopt a startup's software because the UI is pretty. They adopt it because it finds things they missed.
That is also why the human-in-the-loop posture matters more than it sounds. Axion is not pitching automation that shoves the engineer aside. It is pitching augmentation that keeps the engineer in the chair and removes the drudgery underneath them. First hires people who have actually certified aerospace and medical components, maps how they reason, and then teaches the software to handle the repetitive analytical steps one at a time. The expert defines the standard; the machine scales the vigilance.
And the timing is not an accident. First's larger bet is that American manufacturing is entering a phase where the winners are whoever learns fastest from the field. Reshoring, supply-chain anxiety, and the pressure to iterate hardware at software speed all point the same direction. The company that hears its customers' problems first and reshapes the product first will out-run the one that waits for the recall notice. In that world, quality stops being the department nobody visits and becomes the engine of the whole thing.
Joins McKinsey & Company in AI/Strategy. Rises to Engagement Manager, leading consultants and data scientists on C-level analytical bets.
Works inside QuantumBlack, McKinsey's AI arm - machine learning and optimization for automotive and aerospace clients.
Founds Axion (Axion Ray) and becomes CEO. The thesis: stop admiring AI prototypes, start shipping the action.
Lectures on Enterprise AI Strategy at Columbia Business School. Axion takes shape.
Axion raises $7.5M to enhance product integrity at scale.
$17.5M Series A led by Bessemer Venture Partners with RTX Ventures. Total funding hits $25M.
Joins Inspired Capital's Alexa von Tobel on the “Inspired” podcast to talk AI, quality, and the recalls that never happened.
Bachelor's degree.
Master's degree, University of Cambridge.
Master's in Data Science.
Lecturer, Enterprise AI Strategy.
The future of American manufacturing will be defined by how fast you can learn about your customers' needs and iterate your products in response.
Poor quality costs manufacturers trillions every year - and each day unaware of an issue can cost millions more. Axion puts AI directly into the hands of field engineers.
We found two distinct root causes behind surgery equipment failures that were still harming patients, even after the manufacturer thought the problem was solved.
When Axion wins, the world gets safer, innovation gets faster, and everyday products work better for all of us.
If you want the unedited version of the thesis, the man explains it himself. Both worth the time.
► Solving the $4 Trillion Manufacturing Problem with AIInspired with Alexa von Tobel · YouTube ► Daniel First on Axion RayFounder conversation · YouTubeDetect emerging product failures at the earliest possible signal. Kill recalls before they start. Let companies iterate products in real time based on exactly what is breaking in the field - turning quality from a cost center into the sharpest edge a manufacturer has.
Axion was born from watching enterprise AI pilots fail - and deciding the fix was the product.