Founded by Anthropic's first product recruit and METR's former COO, the company subjects agents to approximately 5,000 adverse scenarios and sells the certification that goes with it.
On September 15, AIUC announced a $40 million A series led by Ribbit Capital, with First Harmonic, bringing its total funding to $55 million. The bet of the company, founded in 2024 in San Francisco, is to treat the deployment of AI agents as an insurable risk: to establish an evaluation standard, to have it audited, and then to endorse financial coverage of the result.
This standard, called AIUC-1, presents agents with approximately 5,000 adverse risk scenarios, complemented by quarterly audits to track the emergence of new attack vectors. The business model is that of a trusted infrastructure: the client company does not itself demonstrate the reliability of its agent, it buys a certification that its own customers may require. AIUC indicates that the standard already certifies widely deployed agents, including Cursor, EvenLabs, Fin, Harvey, Livants, UiPath and uses at KPMG.
The profile of the founders weighs heavily on the case: Rune Kvist was the first product recruit from Anthropic, and Rajiv Dattani held the position of operations director of METR, the organization for the evaluation of advanced models. Both therefore come from the environment that measures the dangerous capabilities of models, and transpose this practice to the buyers of agents.
Read the full article on the original source
Go to the source →