Crux Labs, a New York start-up that checks enterprise AI agents against the judgment of a company's own experts, has come out of stealth with more than $6m in funding. The oversubscribed round was co-led by Benchstrength and Misneach, with Analog Ventures and Pareto Holdings taking part. Crux announced the funding on 29 October 2026.
The company did not give the round a stage name or disclose the exact amount above $6m. The site lists it as a seed round. The valuation was not disclosed.
“Aligning AI agents to human judgment is one of the biggest challenges of our time.”Julie Kheyfets, Co-founder and CEO, Crux Labs
What Crux Labs does
Companies deploying AI agents often struggle to know whether the agents are doing the job the way the business wants. Many teams review only a small sample of agent activity by hand, often 1% or less, so most failures go unnoticed, the company says.
Crux uses its own agents to capture expert judgment inside a company's existing workflows, then builds custom evaluations calibrated to that judgment. The company says these evaluations can watch all of an agent's production traffic in real time and feed improvements back into the live system.
Crux says it is already working with Fortune 500 companies and fast-growing start-ups in healthcare, cybersecurity, financial services and other complex industries. It has not named customers or disclosed revenue.
Kheyfets told Unite.AI that agents are very good at reaching goals but lack the rules of thumb that human experts build up over years. Crux's evaluations are used for production monitoring and alerts, and their results feed later improvements to the agents. The company has not said exactly which parts of an agent those improvements change.
Who is behind it
Chief executive Julie Kheyfets and chief technology officer Krishna Kulkarni met at Princeton. Kheyfets previously led the North American business of Tractable, an AI company, and was chief executive of Block Renovation, according to Tech Funding News. Kulkarni led brand engineering at Meta and built AI products at background-check company Checkr.
Kulkarni said engineering teams know how to build AI agents and subject matter experts know what success looks like, but that enterprises lack a bridge between the two. John Monagle, a general partner at Benchstrength, said the companies that win with AI will have agents that reflect the judgment of their best people.
What the money is for
Crux will use the funding to expand its research and engineering teams and to speed up deployments with enterprise customers. Tech Funding News reported that it is hiring researchers and engineers.
The company is entering a growing market for tools that monitor and evaluate AI systems. Tech Funding News cited a projection that the AI observability market will grow from $1.24bn in 2025 to $12.75bn by 2032. Crux's pitch is that general-purpose metrics are not enough, and that each company needs its agents measured against its own standards.
Company website: crux-labs.ai