In most factories, automation is fixed. Robots and machines are programmed for specific tasks and products. Changing a line means reprogramming, physically changing facilities, or replacing equipment, which can take weeks or even months. San Francisco-based Industrial Next Inc. says it’s changing that.
Announced today on the Startup Battlefield stage at TechCrunch Disrupt 2026, the startup is building self-contained AI-driven robotic stations that leverage a unique foundational model that connects vision, task context, and motion. The company says this will allow production lines to be rebuilt in minutes instead of weeks without traditional programming. The goal is to help manufacturers automate complex and delicate tasks that traditionally required human labor.
On stage, the company announced that it is partnering with the BMW Group Technology Office to evaluate Industrial Next’s AI-powered dexterity for manufacturing.
Industrial Next was founded in 2021 by Lucas Pankau and Allen Pan, who previously worked at Tesla. Mr. Pankau served as lead electrical designer at Tesla before moving to Waymo. Pan led Tesla’s self-driving factory efforts and later became chief architect of self-driving company Neolix.
The two came up with the idea for Industrial Next while working together at Tesla. Pan Cow and Pan’s first encounter was when Pan literally tripped over him while he was up late sleeping on the factory catwalk. Although the two didn’t get along at first, they soon realized they shared an obsession with getting things done and had complementary skills.
The founders realized they needed to reinvent manufacturing after years of watching cutting-edge automotive and self-driving technology be hampered by decades-old factory equipment.
“We had all the cool things we could do with cars. We could run machine learning algorithms to design cars. But we couldn’t figure out how to program robots, which was very common for Tesla engineers,” Pankau said. “There was this huge chasm, right? We had people who were very good at what they did, designing cars and designing self-driving systems, and they looked at this equipment from the 1980s in the factory and were essentially stumped. So we started to reimagine what this should look like, using today’s technology, using everything we know today.”
Even after they moved on to their next jobs, they continued to think about their dissatisfaction with industrial equipment. They kept in touch, often remembering machines that didn’t work the way they wanted, and ultimately decided they could do better.
“I was at Waymo at the time, and I missed the pace of Tesla and was kind of bored,” Pankau said. Pan held a similar position at Momenta, a Chinese self-driving system developer, he added. Pan recalled their conclusions. “We’re not getting any younger, and we’re obviously still passionate about it. We still complain about industrial technology, like, this is three years from now, we should start a company.”
Pankau points out that traditional industrial robots are typically programmed to perform very specific and repeatable tasks. As a result, even small process or product changes can cause robot failures and costly downtime, forcing manufacturers to simplify their products to ensure repeatability.
Making robots truly flexible comes with other challenges. Huge amounts of training data are required, but unlike internet-trained AI models, robotics lacks the vast datasets needed to teach machines how to perform complex physical tasks.
“You can make robots do amazing things, like dance, because we have the data to do it. But you can’t make robots build things with high precision,” Pankau explained. “Here we’ve made great progress by using simulated environments to perform these high-precision, high-dexterity tasks. The secret sauce of the robots we use is that we give them an understanding of the general process, but we build that understanding using a very high-fidelity simulation engine that is close to real-life dynamics, allowing us to have the robot perform these kinds of high-precision tasks without ever seeing the parts.”
In other words, the robot practices in a highly detailed virtual world that closely mimics real-world physics. Industrial Next’s robotic stations can be reconfigured, combined and redeployed across the product line. New tasks are taught through simulation and demonstration rather than hard-coded scripts.
Pankau explained that some of the startup’s workstations are already being used to manufacture one of the world’s most popular smartphones. They handle tasks that manufacturers have tried and failed to automate for years, such as threading delicate flex cables and snapping tiny connector caps into place. He also described a “film stripping” station. There, the company’s AI-powered robots can peel protective films from camera modules and buttons with 100% accuracy at faster than human speeds, even when moving to new cell phone models or production lines that the company has never seen before.
“We founded this company because that’s the kind of performance we wanted to achieve with Tesla. We were avoiding robot stations because we couldn’t program them effectively enough using traditional technology,” Pankau said. “Now we want to take this into more complex tasks than just making phones. Eventually, we want to take this into making complete cars end-to-end without any human intervention.”
To check out more of the Startup Battlefield, join us at Disrupt this week in downtown San Francisco, along with a list of prominent founders and VCs with industry commitments you won’t want to miss.
If you buy through links in our articles, we may earn a small commission. This does not affect editorial independence.
