The ideal for venture investors is finding and partnering with a startup that combines a deeply held investment thesis with a strong pre-existing relationship with the founders. Arize AI, which Foundation Capital was the first firm to back and which just announced their Series A, landed right in that sweet spot.
For the better part of a decade, my thesis has been that AI/ML will reinvent the enterprise. I arrived at this viewpoint after more than a decade of experience with machine learning, beginning with deploying it for behavioral targeting when I ran Microsoft’s online ads business and later as an early investor in Conviva, Aggregate Knowledge, and Databricks. At Foundation Capital, my partners and I all share the conviction that we are in the early stages of artificial intelligence and machine learning. While one in 10 enterprises now use 10 or more AI applications, we expect that number to grow dramatically as new technologies emerge that automate and augment existing workflows and human tasks.
As we have expanded AI investments, we have watched many organizations struggle to move from experimentation to implementing AI at scale. At the same time, many more have transformed their businesses entirely on the back of exceptional ML implementations. These scenarios raise the trillion-dollar question: Why do some companies fare so poorly with their ML initiatives while others succeed?
This was the topic Arize CEO Jason Lopetecki and I batted around when he told me he was thinking about starting another company. I’ve known Jason since 2010, when I invested in the prior company that he cofounded, TubeMogul. It was also at TubeMogul where I first met his future Arize cofounder and CPO, Aparna Dhinakaran, when she interned at the company. Our conversations about what kind of startup to work on eventually led to Jason and Aparna making one of the strongest and most compelling cases I’ve heard for the ML-success question: most ML initiatives fail because data teams can’t troubleshoot issues and surface the “why” behind the complex decisions being made by their ML models in production.
Their argument, born from their experience at some of the world’s most innovative, data-driven companies, was so persuasive that we led the company’s first round of funding and are proud to be participating in their Series A.
Today, Arize AI has built the only Machine Learning Observability platform that helps ML practitioners successfully take models from research to production with ease. Arize’s automated model monitoring and analytics platform help ML teams quickly detect issues when they emerge, troubleshoot why they happened, and improve overall model performance. By connecting offline training and validation datasets to online production data in a central inference store, ML teams can streamline model validation, drift detection, data quality checks, and model performance management.
For the past 18 months, we’ve watched the Arize vision become reality as AI has continued to fuel business functions at nearly every kind of business. Even more critically, an unexpected, extreme outlier event created massive ripples throughout the systems that run the world. In an instant, the models driving every single ML implementation — from healthcare and finance to the gig-economy, credit, commerce, and auto-traffic — became obsolete. As a result, demand for model monitoring and observability services has surged as organizations have been forced to adapt and retool their systems in response to the global pandemic.
The pandemic has created a new world and companies that rely on AI have learned far-reaching, long-term lessons. Most notably, companies that are putting ML models into production without the ability to monitor, explain, and troubleshoot issues are leaving critical parts of their business up to chance. This is a fundamental shift that has transformed the ML infrastructure landscape faster than we could have ever predicted. This is evidenced by the speed with which Arize has grown its customer base. Companies like Adobe and Twilio are two examples of the many organizations that have tapped Arize to keep their ML initiatives on track in an increasingly unpredictable world.
As their first investor, I’m thrilled about the progress that Arize has made since our initial investment. I’m equally inspired by the fact that the company is executing its Day-One vision to build solutions that act as the ethical guardrail on AI as it’s deployed in the real world. Now more than ever, providing transparency and introspection into historically black box systems can ensure more effective and responsible AI, and Arize is doing just that.
When a company can continuously improve the performance of incredibly complex technology systems while positively addressing bias and ethical issues that are impacting large parts of society, the future is incredibly bright. I often wonder if the early days of AI would have turned out differently if data teams better understood why machines, trained by data, made the decisions they did and how to get them back on track when something went wrong. Arize finally gives us the opportunity to find out. It’s an ambitious mission, and one we are proud to be behind.