Arize AI builds an observability platform to evaluate AI models and monitor them pre- and post-launch across use cases from machine learning and computer vision to generative AI and agents. The company uses a “council of judges” approach that can include multiple AI models and humans in the loop to detect and diagnose failures. Arize was founded by Jason Lopatecki and Aparna Dhinakaran in 2019 out of experiences at TubeMogul (acquired by Adobe for over $500 million in 2016) and is based in Berkeley, California. Its platform is used by enterprises including Uber, Klaviyo, and Tripadvisor, and the company also publishes an open source offering, Arize Phoenix, which has more than two million monthly downloads. Arize reports explosive growth over the past two years as AI adoption accelerated. The company plans to use new funding to improve its core product, double down on growing AI segments such as voice and AI agents, and further invest in its open source Phoenix project. Arize AI provides a machine learning operations (MLOps) platform that emphasizes observability for production models, including tools to inspect embeddings, trace bias, and monitor embedding drift. The company offers features like an embeddings product, a “Bias Tracing” tool, and embedding drift monitoring to surface when models degrade due to outdated training data. Arize counts customers including Uber, Spotify, eBay, Etsy, Instacart, P&G, TransUnion, Nextdoor, Stitch Fix and Chick-fil-A, and its free tier launched earlier this year has over 1,000 users. Founded in 2019 by Jason Lopatecki and Aparna Dhinakaran, the company says it aims to help teams understand and troubleshoot models in production rather than build observability tooling themselves. Financially, Arize raised a $38 million Series B that brings total capital raised to $62 million, and CEO Jason Lopatecki said the proceeds provide ample runway. The company plans to scale R&D and double its current 50-person headcount over the next year. Arize AI offers an ML observability platform that monitors, explains and troubleshoots model and data issues and can be integrated into customers' AI systems within 30 days. The company targets deployments across fintech, healthcare, insurtech, adtech and retail and cites enterprise customers including Adobe and Twilio. Founders Jason Lopatecki and Aparna Dhinakaran bring prior experience building ML infrastructure and observability tooling. Arize has scaled from product development to “massively deployed” usage and grew its team to about 40 employees with plans to double headcount. Management intends to double down on product development, expand industries and use cases, and put the new capital toward hiring. The company reports 100% annual recurring revenue growth each quarter and a doubling of its customer base. Arize AI is developing a real-time analytics platform focused on production observability for AI and ML models. The product surfaces real-time analytics and insights to data scientists, product managers, and engineering teams to help determine whether issues stem from data or the model itself. The company positions the tool as an engineering product—likened to Google Analytics for models and similar to Splunk—rather than an executive dashboard. Arize launched out of stealth and has already acquired YC-backed Monitor ML; the entire Monitor ML team joined Arize and its CEO Aparna Dhinakaran became co-founder and chief product officer. The company is led by CEO Jason Lopatecki (formerly of TubeMogul) and lists TubeMogul co-founder Brett Wilson as an investor and board member. Arize has raised $4 million in funding led by Foundation Capital, with participation from Brett Wilson and Trinity Ventures, and is focused on making it easier to monitor, troubleshoot, and scale deployed models so senior researchers spend more time on research and less on firefighting.

Total raised

$131M

Funding rounds

4

Latest round

Series C

Latest activity

Feb 2025