ChatGPT's public release in 2022 rippled through the professional services world. Entire industries like law and accounting quickly understood that artificial intelligence would change the way they did business.

Now the startup Noetive wants to create a similar inflection point for companies whose business is done not on a computer, but on the ground.

The fledgling AI lab is developing an industrial AI model for businesses in physical industries, focusing for now on manufacturing, logistics and construction. It emerged from stealth Wednesday, announcing a $41 million seed funding round led by venture firm Eclipse.

The company is helmed by Eclipse's chief AI officer, Amir Frenkel, and Dan Barak, who was most recently vice president of product at web development platform Netlify.

"I can ask ChatGPT: What does it feel like to ride a bike? I don't think I would ask it to ride a bike for me," said Frenkel, who previously served as vice president of GenAI for Meta. "It has no or very limited understanding of the 3-D, or really 4-D, space."

It isn't just ChatGPT. Large language models on the whole have limited capacity to operate in the three-dimensional world. But then there are world models-AI built to understand physical environments. Several major labs developing the technology have launched in recent years, including Stanford Professor Fei-Fei Li's World Labs and former Meta Chief AI Scientist Yann LeCun's Advanced Machine Intelligence Labs. Both announced billion-dollar fundraises earlier this year.

The models developed by these labs could help robots, vehicles and other devices and machinery operate autonomously in unpredictable, real-world environments. Investment in physical AI-things like autonomous vehicles and intelligent robotics-is on track to rise from $13 billion in 2025 to $47 billion in 2026, according to a July report from Silicon Valley Bank, a division of First Citizens Bank, citing data from PitchBook.

Noetive's aim is narrower: developing a world model purpose-built for the three sectors it is currently focusing on.

Its platform will analyze businesses' operations by sitting atop existing software used by customers-systems that are often outdated and highly custom, which makes them difficult to integrate with generalist AI models, said Eclipse co-founder Lior Susan. Then, drawing on its model's understanding of the physical world, Noetive will help businesses improve their processes.

Elma, N.Y.-based beverage manufacturer Steuben Foods is now using Noetive's platform to compile its weekly manufacturing schedule, said co-CEO Menachem Katz. Steuben is a contract manufacturer and has its own brand of beverages. Katz described the creation of the schedule for the company's production lines as a complex exercise that must juggle a number of variables, including supply-chain disruptions, vendor and customer behavior, and employee availability.

Working with Noetive, Katz said, Steuben is on a path to increase the capacity of its existing production lines by between 10% and 15%.

"The way that their system is capable of taking everything into consideration when creating a schedule, that's one of the things that-you're immediately unlocking more capacity," he said. "That means, immediately, more customers that can produce, and more team members that can join the company."

Noetive declined to disclose the valuation at which it raised the seed round, which included participation from Craft Ventures and the Westly Group, among others. Angel investors, including Meta Chief Technology Officer Andrew Bosworth and Airbnb Chief Technology Officer Ahmed Al-Dahle, also participated.

Susan said he approached Frenkel with the idea for Noetive early this year.

"We all ran physical operations in our operating life, we all struggle with this clunky software that is on-prem, that is highly custom," Susan said of the team at Eclipse. The firm, which has about $12.5 billion under management, describes itself as operators with capital.

"We're like, 'Hey, this is a huge problem. No one has solved it yet. It's really tough; it's really technical,'" Susan said, describing his pitch to Frenkel. "And he just got obsessed."

The seed funding will go toward hiring as well as research and development. The latter includes the training of Noetive's AI model, which is recursive, according to Frenkel and Barak, or designed to improve itself over time. It also includes the development of a universal sensing pod, which will provide real-time spatial insights to businesses-observing the state of their supply lines or fleet of trucks, for example.