Most factory robots are rigid. They follow fixed paths, require precise part placement, and need reprogramming every time something changes on the line. Inbolt is solving exactly that problem. The French startup builds physical AI and 3D vision systems that let industrial robots perceive their environment and adapt their motion in real time, without manual teaching or mechanical redesign. In 2025 alone, Inbolt’s systems powered 40 million robot cycles across its deployed base. The company is already deployed in more than 100 factories, working with manufacturers like Ford, Stellantis, Toyota, and thyssenkrupp. Its Series A was led by Exor Ventures, the investment arm of the same group that holds major stakes in Ferrari and Stellantis, so some of its biggest backers are also its customers.
The company was co-founded in 2019 by Rudy Cohen, Albane Dersy, and Louis Dumas, who met during the MSc X-HEC Entrepreneurs program. With backgrounds in mathematics, AI, and mergers and acquisitions, the team identified the industrial automation problem during an exchange at UC Berkeley, where they saw how inflexible most robotic systems really were. Cohen leads as CEO, Dumas as CTO, and Dersy as COO.
The Technology Behind it:
Inbolt’s platform has two core parts: hardware and software working together. The hardware is a compact 3D vision camera that mounts directly on the robot arm. It captures detailed spatial data even in challenging factory environments, with no need for jigs or rigid setups.
The software side is where the real work happens. Inbolt’s proprietary AI model processes that visual data and calculates optimal robot paths in under 80 milliseconds, from perception to motion. The system trains using only a CAD model of the part, so deployment takes minutes rather than weeks. The configuration tool, called Inbolt Studio, is built for factory engineers and integrators, not just AI specialists.
24 Hours to Full Deployment:
One of the more practical things about Inbolt is the deployment speed. The company says a full station can be up and running within 24 hours. ROI payback lands around six months, which is fast for capital equipment in manufacturing.
The system is compatible with any major robot brand, so factories don’t need to replace existing hardware. The vision unit retrofits onto current robots, and the AI model takes over guidance from there. For new stations, the same system comes pre-integrated from day one. Inbolt also supports moving production lines, which is a harder problem that most vision systems avoid entirely.
A Real Bin Picking Problem, Solved:
One of the cleaner examples of what Inbolt actually does is the work with LAN Systems at a Stellantis plant. The task was bin picking crankshafts, parts that come in varying sizes and land in bins without any consistent orientation. That combination of irregular bins and variable part geometry is exactly what stops fixed-path robots cold.
Inbolt’s 3D vision system handled it without redesigning the station or adding mechanical constraints. The Technical Director at LAN Systems noted that automating the process had been impossible before Inbolt, specifically because of the irregular bins and varying crankshaft sizes. That’s a concrete use case, not a demo reel.
Real Results From the Factory Floor:
The customer feedback here is specific enough to be credible. A production lead at thyssenkrupp noted that non-compliant tightening dropped below 1% after deploying Inbolt’s 3D vision guidance. A maintenance manager at Stellantis described how retrofitting a high-risk station with vision turned it into a reliable automated cell, with no mechanical re-engineering required, and the project is now expanding to similar cells across the plant. An initial contract with Stellantis alone enabled the manufacturer to save more than €3.1 million in one year.
Partnerships and Program Recognitions:
Inbolt has built a notable set of industry partnerships. Its AI software is integrated into the NVIDIA-powered Universal Robots AI Accelerator Kit, bringing vision-guided robotics to a wider base of UR installations. The company is also part of the NVIDIA Inception program and has collaborated with FANUC on moving assembly line automation, which involves robots tracking and working on parts that are in motion.
In late 2025, Inbolt was selected as one of 80 companies in France’s Tech 2030 program, a government-backed initiative focused on scaling high-impact technology companies. The company has raised a total of approximately $22 million across three funding rounds. Inbolt now has offices in France, the US, and Japan, and is actively deploying across automotive, electronics, and home goods manufacturers on three continents.
