Micro1 Partners With Box for Agent Evals
AI agent platform Micro1 has partnered with cloud content management company Box. The collaboration will focus on human-grounded evaluation for AI agents operating in production environments. The goal is to validate agent behavior when interacting with sensitive enterprise documents.
- Micro1, which provides human experts for AI data labeling and model training, raised a $35 million Series A in September 2025 at a $500 million valuation. The company reported its annual recurring revenue grew from $7 million to over $50 million in 2025. - The company is a key competitor to Scale AI, gaining market share after some major AI labs sought alternatives due to Scale's deepening relationship with Meta. - Box's strategy centers on its Box AI platform, which integrates models from providers like OpenAI to allow enterprises to analyze and generate content from their own secure data. This partnership with Micro1 provides a mechanism to validate the behavior of these AI agents on sensitive corporate information. - Human-grounded evaluation complements automated and model-based evaluations by using human experts to assess complex agent behaviors that are difficult to measure, such as contextual accuracy, safety, and hallucination. - Micro1's core product is an AI recruiting agent named "Zara," which sources and vets the human experts that form the foundation of its data services. - The collaboration aligns with Box's recent introduction of Box AI Agents, a service focused on automating content-centric workflows and turning unstructured data into "governed, AI-ready assets". - Box's Director of Product Management for AI agents, Matt Terrell, has stated that a key part of the company's AI strategy is transforming enterprise content into context that can be used by various AI tools.
