Detroit’s Inbolt Launches Vision-enabled Robot Programming

Inbolt, the Detroit-based robot intelligence company, is launching two new capabilities that complete the company’s AI vision model for robot guidance: Inbolt Robot Programming and an expanded Inbolt Robot Control.

Both technologies will debut at the Automate 2026 trade show at Chicago’s McCormick Place from June 22-25.

“Robot deployment still takes weeks because the digital twin never matches the real factory floor, engineers hand-tune every trajectory during commissioning,” says Rudy Cohen, co-founder and CEO of Inbolt. “With Robot Programming, the Vision Model, and Robot Control on a single platform, that gap closes.

“Engineers build the program from the CAD, our vision model locates the real part, and the robot executes the planned path. One platform from perception to motion, on the robots manufacturers already own. That’s AI perception built for the factory floor.”

With Robot Programming and Robot Control, Inbolt says it covers the full path from virtual commissioning to adaptive robot motion control, for stationary and moving-line applications.

Until now, the company says, deploying a robot on a factory floor took weeks as engineers carefully build digital twins of the production line, then spent the commissioning window touching up trajectories point by point because the virtual environment never fully matches reality. If the robot is anchored 2mm off, or parts arrive in unrepeatable positions, every path gets re-taught and tuned by hand.

With the latest release of Inbolt Robot Programming, the programming capability inside Inbolt Studio removes that step entirely. Engineers build the program directly on the CAD model, in the part’s own reference frame. At runtime, the Inbolt Vision Model locates the real part and adjusts the robot’s motion to execute the planned path exactly.

“No teach pendant. No iterative tuning. No separate workflow for moving lines,” says Cohen. “Weeks of commissioning now works in one shot. The digital twin and the factory floor are the same thing.”

The CAD-based release is available for FANUC, Universal Robots, and Yaskawa on dynamic (moving line) applications, with broader brand coverage on the roadmap. Two of Inbolt’s four Automate 2026 booth demonstrations will run it live, so visitors can watch the system go from CAD to executable robot motion in front of them.

“Automate in Chicago is where we plant our flag in the U.S.,” says Albane Dersy, co-founder and COO of Inbolt. “Four live demos, two product launches, a deep integration with FANUC and NVIDIA on the show floor, and a panel on the future of physical AI. Our U.S. footprint has expanded across Stellantis, GM, and Toyota plants this year, our team has doubled, and the U.S. contingent doubles again by year-end.”

Inbolt’s second product release is an expansion of Robot Control, the real-time robot motion execution component of the platform, now running natively on Yaskawa, joining FANUC, KUKA, ABB, Universal Robots, and Comau.

Robot Control streams corrected joint commands directly into the robot’s servo loop at native control frequency, closing the loop between what the vision model sees and how the robot moves. The Yaskawa expansion brings Inbolt’s native robot brand coverage to six, giving manufacturers a single intelligence layer for real-time execution across the brands they already own.

Inbolt also has released updates to the Inbolt Vision Model with improved global part localization models. The model now tracks a wider variety of parts, and the Inbolt Studio dashboard exposes part position, detection status, and live performance tests for each use case. Robotics engineers can troubleshoot and evaluate Inbolt’s performance on their specific station inside Inbolt Studio.