Arize

About Arize

Arize AI positions itself as a critical infrastructure provider in the artificial intelligence and machine learning sector, focusing on ML observability and evaluation. The company was founded in 2019 by Jason Lopatecki (CEO) and Aparna Dhinakaran (CPO) and officially launched in 2020. The founders' journey stems from personal frustrations with the opaqueness of production ML models. Lopatecki, a "garage-to-IPO executive," previously co-founded and served as Chief Innovation Officer at TubeMogul, bringing extensive experience in analytics and scaling data-heavy businesses. Dhinakaran has a background as an ML engineer at prominent tech firms like Uber, Apple, and also TubeMogul, where she worked on core ML infrastructure platforms, including Uber's Michelangelo. Their combined experience highlighted the difficulty in troubleshooting AI when it malfunctions, which directly informed Arize's mission.

The firm's core business is a SaaS platform that allows AI developers and data scientists to monitor, analyze, and troubleshoot deployed AI models in real-time. This addresses the significant challenge of models behaving differently in the real world compared to their performance in a research environment. Arize serves a broad range of clients, from startups to large enterprises like Condé Nast, Discord, and Priceline, across various industries including finance, e-commerce, and fleet management. The business model is primarily subscription-based, offering different tiers of its platform, including a self-serve option to broaden adoption.

The Arize platform offers a unified solution for both traditional ML models (like classification and regression) and modern generative AI, including Large Language Models (LLMs). Key features include automated monitoring for performance degradation, data drift, and data quality issues. Its performance tracing capability helps users pinpoint the root cause of problems by slicing data to identify problematic cohorts. For the burgeoning generative AI space, Arize provides specialized tools for LLM and RAG (Retrieval-Augmented Generation) system evaluation, prompt optimization, and tracing. A notable product is Phoenix, an open-source library for LLM evaluation and observability, which has gained significant traction. The platform is built on open standards like OpenTelemetry, ensuring interoperability with a client's existing tech stack.

Financially, Arize has demonstrated significant investor confidence, having raised a total of $131 million over four funding rounds. Its most recent funding was a $70 million Series C round in February 2025, led by Adams Street Partners with participation from firms like M12 (Microsoft's venture fund), TCV, and Battery Ventures. This substantial investment, noted as the largest in AI observability, is aimed at scaling its evaluation capabilities for advanced AI systems and expanding partnerships with major cloud providers like Microsoft and Google.