Investors
Repeatable high-mix deployment at software economics.
Cognivix builds the AI layer that lets an existing industrial robot learn a new assembly task from a single human demonstration. Every new part today costs about five engineer-days of programming, fixturing and debugging. AiKit compiles the same task in about five minutes, with no robot code.
Thesis
Why now
- Most manufacturing tasks are high-mix and still manual. Traditional automation was built for high-volume, stable production: every new part means programming, custom fixtures and weeks of integration, costs that erase the benefit when batches are small.
- Labor shortages and reshoring are forcing factories in the US and Europe to automate work that was never economical to automate.
- Foundation-model robotics needs thousands of teleoperated demonstrations per task and a trained policy per SKU. AiKit needs one demonstration and compiles an explicit, inspectable task graph that transfers to new poses, fixtures, robots and grippers.
The open market is not robotics. It is repeatable high-mix deployment at software economics: a recurring subscription per line, deployed by certified integrators on an installed base of millions of arms.
Where we are
Available now, first pilots starting
Product
- AiKit available now: retrofit kit and software for one arm, $10k per line per year, hardware included
- FANUC and ROS 2 arms supported at launch
- AiCell (two-arm coordination, complex assembly) demonstrated in the lab, targeted for 2027
Commercial
- IDT Solution certified as first AiKit systems integrator for the US and EU (July 2026)
- Active evaluations with Fortune 500 manufacturers and with NTT Data
- Member of the NVIDIA Inception program
- Operating from San Francisco and Pescara, Italy
Team
Founders and founding team
Founders
Daniele Bernardini, PhD
Co-founder and CEO
Architecture and task compiler. 25 years at Siemens, OSRAM and TUM.
Lukas Dirnberger, PhD
Co-founder and CTO
Robotics ML and prototypes. TUM.
Giorgio Calapa
Co-founder and COO
Operations and delivery. 20 years in fintech and investment banking at Bloomberg and then independently, with clients including Bank of America and Citi.
Francesco Scalise
Co-founder and CRO
Manufacturing commercialization. 25 years in manufacturing, with clients including Caterpillar and Lamborghini.
Founding team
Prof. Boyoung An, PhD
Founding Scientist
Perception and robotics.
Suraj Deshini, MSc
Founding Engineer
Robot execution and controls. Industrial PhD with TUM.
Silvia Trovato
Go-to-market
Rollout and program management.
The task compiler grows out of the founding team’s published research with the Technical University of Munich and UC Berkeley.
- Learning to Generate All Feasible Actions (IEEE)
- 6IMPOSE: bridging the reality gap in 6D pose estimation for robotic grasping
- Exploring Deep Learning Grasp Models with Soft Anthropomorphic Robotic Hands
Institutional investors
Get in touch
We share the full deck, the uncut demonstration footage and pilot data under NDA with robotics and physical AI investors.
Community round
Cognivix is also running a community investment round on Republic, open to customers, partners and supporters who want to hold a stake in the company alongside institutional investors.
Investing involves risk. Please review the full offering materials on Republic.
© 2026 NVIDIA, the NVIDIA logo, and NVIDIA Inception are trademarks and/or registered trademarks of NVIDIA Corporation in the U.S. and other countries.
