The Lawyer Who Went to the Factory Floor
He had the corner office. He had the title. He had 75 attorneys reporting to him at one of the defining companies of the internet era. Then Jon Sobel walked away from all of it to spend a year asking factory managers why their data was useless.
That question - asked 40 or 50 times, in plants across America - became Sight Machine. The answer was always the same: manufacturers were drowning in sensor readings, machine logs, and production metrics they had no way to actually use. The data existed. The insight did not. Nobody had built the abstraction layer to make it work.
Sobel saw the gap. It took an unusual career to see it clearly.
Start at Princeton, studying public and international affairs. Then law school at Michigan, graduating in 1990. A brief stint as an actual newspaper reporter in Michigan. A policy adviser role in Washington D.C. to Senator John Danforth of Missouri. Then a San Francisco law firm. Then, in 1998, Yahoo - where Sobel joined what was still a scrappy internet company and watched it become a global phenomenon, running a legal team that grew to more than 75 attorneys as SVP, General Counsel, Board Secretary, and member of the executive committee.
After Yahoo: CBS Interactive, where he worked on online video and mobile strategy. Then SourceForge and Slashdot, managing 30 terabytes of data per day and 2 million open-source software files distributed globally. The open-source world taught him something about meritocracy and non-hierarchical collaboration that stayed with him. Then Tesla - where he joined as General Counsel in September 2009, when Elon Musk's electric car company had fewer than a thousand employees and the Roadster was barely in production. Then Calera, a carbon capture startup.
None of those detours were wasted. At SourceForge, he learned what industrial-scale data infrastructure feels like from the inside. At Tesla, he watched software transform a physical product industry. At CBS, he saw what happens when legacy industries try to adapt to data-driven change and mostly fail to.
Manufacturing is the NFL of data variety - and no one was calling the plays.
- Jon Sobel, Sight MachineThe Insight Nobody Wanted to Have
Sobel co-founded Sight Machine in 2011, starting in Michigan - close to the industrial heartland, not in the startup bubble. The initial insight was uncomfortable: factories generated enormous amounts of data, but that data was structurally unusable without a purpose-built platform to make sense of it. Every plant was different. Every machine spoke a different protocol. Every production run generated a different shape of output.
"Manufacturing is the NFL of data variety," Sobel has said. The analogy is precise. Just as football requires coordinating players with different roles, speeds, and signals into a coherent system, factory data requires an abstraction layer that can normalize radically heterogeneous inputs into something a data scientist - or an AI model - can actually work with.
The prescription he heard from Silicon Valley was naive: just put AI on it. Sobel was not naive. "What we cannot do today - and what no one should expect - is just take a whole big blob of factory data and put AI on it," he said. "It doesn't work. There's no underlying structure that AI can work on." This was not pessimism. It was a product requirement. Sight Machine's entire architecture is built around creating that structure first.
The company raised over $124 million over the following decade, attracted blue-chip industrial customers across 20+ industries and 20+ countries, and became what Sobel set out to build: the category leader for manufacturing analytics.
The Signal in the Noise
"The key to successful digital transformation is to pick projects appropriate for the readiness level of a company or plant."
"Organizational factors are at least as important as technical factors. At least as important."
"Technology is a means to an end. Let's really focus on value."
"When 20 percent of an industry gets serious, that's a tipping point."
"Data scientists spend the majority of their time manually selecting, cleansing and combining data - not analyzing data."
"Manufacturing has long been seen as one of the best opportunities for AI, but the industry has been held back by challenging data environments in plants."
What 14 Years of Building Looks Like
Total funding raised, including equity from NVIDIA NVentures (2025)
Annual revenue with a 65-person team - exceptional revenue efficiency
Countries where Sight Machine's platform operates
Cycle time improvement at an early client - "an enormous leap for a highly automated company"
From Newsroom to Factory Floor
The Details That Explain the Mission
The Factory Tour: Before writing a line of product code, Sobel visited 40 to 50 manufacturing facilities. He heard the same pain points at every single one - data they couldn't use, insights they couldn't act on, analytics that required a data science PhD to touch. The repetition was the signal.
The 7% Moment: One of Sight Machine's first clients discovered unexpected cycle time inefficiencies through the platform. A hackathon with floor operators yielded a 7% improvement within weeks. "An enormous leap for a highly automated company," Sobel notes. Seven percent doesn't sound like much until you run the math on a facility running 24/7.
The SourceForge Lesson: Managing 30 terabytes per day and 2 million software files at SourceForge wasn't just a data challenge - it was a lesson in meritocratic, non-hierarchical collaboration. Open source's flat power structures became part of Sight Machine's company DNA.
The Journalism Fellowship: Sobel holds a fellowship from the Poynter Institute for Media Studies - one of journalism's most respected training programs. For a manufacturing AI CEO, it's an unusual credential. It explains why he communicates with unusual clarity about a technically complex domain.
By melding rapidly advancing AI technologies with the unique semantic layer generated by Sight Machine's industrial platform, we have bridged the gap between general-purpose AI and the extremely challenging nature of production data.
- Jon Sobel, September 2025What Comes Next
In September 2025, Sight Machine announced its industrial AI agents - tools that let operators without data engineering expertise use complex plant data to improve production. The announcement came with an equity investment from NVentures, NVIDIA's venture arm, and expanded integration with Microsoft Fabric and NVIDIA Omniverse. In 2026, the company is previewing AI agent crews at Hannover Messe - systems that can simulate possibilities, recommend optimal settings, and progressively take direct control of specific factory parameters as they earn reliability trust.
Sobel's aspiration is not subtle: comprehensive, practical AI in every manufacturing plant in the world. Not AI as press release. AI as the thing that keeps a line running at 3 a.m. when the shift supervisor is three decisions away from a shutdown.
He also teaches. As an adjunct professor at the University of Michigan, Sobel runs graduate courses in innovation, entrepreneurial leadership, and solving complex problems. The students get the version of Silicon Valley's relationship with industrial America that most Silicon Valley executives never bother to explain - which is that the factories making the things people buy are far more complex, far more data-rich, and far more underserved by software than anyone outside of them realizes.
The lawyer became the entrepreneur. The entrepreneur became the teacher. The teacher is still building.
Verified Milestones
- Co-founded Sight Machine in 2011; grew to category leader in industrial AI / manufacturing analytics
- Raised $124M+ in total funding including equity from NVIDIA's NVentures (2025)
- Sight Machine platform deployed across 20+ industries and 20+ countries
- Led Yahoo!'s legal department as SVP & General Counsel; team grew to 75+ attorneys
- General Counsel at Tesla Motors during critical pre-IPO growth period (2009)
- Sight Machine named finalist for 2025 Microsoft Manufacturing Partner of the Year
- Achieved ~$21M annual revenue with 65-person team - top-quartile revenue-per-head for enterprise SaaS
- Adjunct Professor at University of Michigan teaching innovation and entrepreneurial leadership
- Authored "Manufacturers Struggle to Turn Data Into Insight" (2014) - still cited in manufacturing analytics discourse
How He Thinks
Sobel is not a hype merchant. In an industry drowning in "AI-powered" claims, he is the one explaining at conferences why most AI implementations in factories fail - because the data is not structured, the organizational buy-in is absent, and the project scope does not match the company's readiness level. He recommends books about fighter pilots and B2B subscription economics. He talks about open-source meritocracy as a management philosophy.
His recommendation list includes Boyd: The Fighter Pilot Who Changed the Art of War - a biography about developing decision frameworks under radical uncertainty. The selection reveals something: Sobel is thinking about strategy at the speed of changing reality, not just at the speed of product roadmaps.
Four Disciplines, One Mind
-
PPrinceton UniversityA.B. in Public and International Affairs / Public Policy
-
MUniversity of Michigan Law SchoolJuris Doctor (J.D.), 1990
-
WThe Wharton School, University of PennsylvaniaMaster of Business Administration (MBA), 2007
-
JPoynter Institute for Media StudiesJournalism Fellowship