Analysis
Crux Labs has emerged from stealth with more than $6 million in oversubscribed funding to develop technology that helps enterprises evaluate AI agents against human expertise, the company said, with the round co-led by Benchstrength and Misneach, according to the company's announcement.
What Crux Labs does
Crux Labs helps enterprises capture judgment from their own existing workflows, build custom evaluations, monitor production AI-agent traffic, and use the results to improve agents before pushing updates live. The company said it is already gaining traction with Fortune 500 companies and high-growth businesses in healthcare, cybersecurity and financial services. Co-founder and CEO Julie Kheyfets previously led North American operations for Tractable AI and served as CEO of Block Renovation; co-founder and CTO Krishna Kulkarni previously led brand engineering at Meta, worked on AI products at Checkr, and helped scale Healthify to roughly 20 million patients. The two met at Princeton. "Crux helps every enterprise to distill this judgment and to use it to rapidly improve their agents," Kheyfets said, while Kulkarni described the company's approach as operationalizing "domain expertise into continuous agent improvement."
Crux Labs funding history
The more than $6 million round is Crux Labs' first disclosed financing since the company came out of stealth. The company has not stated a round type, valuation, or any earlier funding.
“The company said it is already gaining traction with Fortune 500 companies and high-growth businesses in healthcare, cybersecurity and financial services.”
Who invested in Crux Labs
Benchstrength and Misneach co-led the round, with Analog Ventures and Pareto Holdings participating. John Monagle, general partner at Benchstrength, said "the companies that win with AI will have agents that reflect the judgment of their best people," framing the firm's investment thesis around the same human-in-the-loop premise Crux Labs is building its product on.
AI agent evaluation market and competitors
The human-judgment-for-AI category drew widespread attention after Meta's reported $15 billion stake in Scale AI in early 2025, which reshaped how frontier AI labs think about sourcing human expertise for training and evaluation data. Since then, competitors including Outlier, Mercor and Invisible have built businesses paying expert evaluators between $50 and $120 an hour, while Surge AI, previously bootstrapped, is reportedly exploring a $1 billion raise to scale its own infrastructure and compliance capacity. Newer entrants Sepal AI and Micro1 are also gaining traction with frontier labs, according to a report on the sector. Pulse's own funding tracker shows AI-evaluation and data-quality startups have been one of the more consistently funded enterprise-AI subcategories this year, alongside cloud-cost tooling and agent observability. Crux Labs' pitch differs from most of these in audience: rather than selling labeling or evaluation services to frontier AI labs themselves, it is selling agent-evaluation tooling directly to enterprise customers deploying third-party AI agents, a downstream market that has fewer established players than the lab-facing data-labeling business Scale AI and its peers built.
The risk for Crux Labs is that its enterprise-facing niche could still be commoditized the same way lab-facing data labeling has been, if Outlier, Mercor or Invisible decide to move downstream into enterprise agent evaluation themselves rather than staying focused on frontier-lab customers.
What to watch next
Watch whether Crux Labs discloses a round type and valuation as it moves past this initial raise, and whether its enterprise customer list grows beyond the unnamed Fortune 500 logos cited in the announcement -- named reference customers would be the clearest signal the Fortune 500 traction claim is durable rather than a handful of pilot engagements.