Covariant

Founded as an academic research project by UC Berkeley and Open AI, Covariant is an AI robotics startup that claims to be building a universal AI (called the Covariant Brain) for robots to be capable of seeing, reasoning, and acting autonomously. It provides robotic gripping software based on reinforcement learning AI techniques. Covariant has stated that the software will enable warehouse robots to pick objects faster than human workers, with approximately 95% accuracy.  

Although founded in 2017, Covariant launched from stealth in January 2020, by which time its platform was already deployed and fully operational at facilities in the pharmaceutical, electronics, and apparel industries, particularly in North America and Europe. By July 2021, the Covariant Brain solution was deployed in warehouses across Europe, North America, and Asia-Pacific. Use cases spanned several industries, including industrial supply, health and beauty, fashion, pharmaceuticals, grocery, parcel, and general merchandise.

In March 2024, the company introduced RFM-1, a robotics foundation model that enables users to build generalized AI models that can simulate and operate in real-world conditions.

Key customers and partnerships 

To expand its customer base internationally and benefit from synergies, Covariant has partnered with companies such as robotics giant ABB (February 2020) and warehouse logistics technology provider Knapp (March 2020). 

Funding and financials 

The company raised USD 75 million in Series C funding in April 2023, co-led by Radical Ventures and Index Ventures, to invest in further developing its universal Covariant Brain, to cater to more use cases in a variety of sectors.