Arize AI is an AI engineering and observability platform built to help teams develop, evaluate and monitor large language model applications, AI agents and traditional machine learning models in production. The platform spans tracing, evaluation, prompt management, dataset curation and live monitoring, with the goal of closing the loop between AI development and real-world performance.
Core capabilities include LLM tracing through OpenInference and OpenTelemetry, automated and human-in-the-loop evaluation, drift and data quality detection, and structured workflows for debugging agent failures. Arize also maintains Phoenix, a popular open-source observability library widely used by AI engineering teams as a lightweight entry point to its commercial platform.
Arize was co-founded by Jason Lopatecki (CEO), who previously led AI strategy at TubeMogul, and Aparna Dhinakaran (Chief Product Officer), formerly of Uber's ML platform team. The company is headquartered in Berkeley, California, and serves enterprise customers across financial services, healthcare, retail and technology.
In February 2025, Arize closed a $70 million Series C led by Adams Street Partners, with participation from M12 (Microsoft's venture fund), Sinewave Ventures, OMERS Ventures, Datadog, PagerDuty, Industry Ventures and Archerman Capital. Existing investors including Foundation Capital, Battery Ventures, TCV and Swift Ventures also participated. The round is among the largest single financings in the AI observability category.
Arize differentiates from competitors such as LangSmith, WhyLabs and Datadog's LLM observability by combining a strong open-source community around Phoenix, deep evaluation tooling for agentic workflows, and enterprise-grade deployment options. The platform is positioned as a core layer of the AI engineering stack for teams operating LLM and agent applications at scale.