Why We Exist
We are a team of people extremely passionate about the potential of AI and Machine Learning.
Who We Are
Our team hails from Berkeley EECS, Uber ML infrastructure, Google & Facebook Engineering and TubeMogul/Adobe real time analytics/statistics team. The team has lived and experienced first hand, scaled businesses running thousands of models built nightly.
Arize AI was started by building on an insight from our own experiences as data scientists, engineers and product leaders. We spent an immense amount of time and energy in data prep and model building, only to have a model thrown into production with diffuse ownership and a hanging question around everyones’ head, is the model even working??!
Launching AI into the Real World
The promise of a future with AI touching every product and business you interact with is exciting to some and scary to others. For some of us deep in the industry, it makes us nervous to think that the current systems and pipelines churning out models, are producing models that make life and death decisions. These models are deciding loan rates from historically biased datasets and deciding when self-driving cars are applying their brakes. They make decisions that can lose a business millions of dollars in a day and generate painful experiences for customers, that very few can troubleshoot.
A future without software that can watch, troubleshoot, explain and provide guardrails on AI, as it is deployed into the real world, is a scary one.
The ML space can be confusing and even understanding what some companies do feels like you need a PhD. Arize AI is laser focused on Production ML. Helping companies stand up their models in the real world. Our AI Observability platform helps companies figure out what is wrong, where it is wrong and why.
In my last business, Brett the Ex-CEO of TubeMogul said the following, which I think captures the state of the industry “I’m spending $10 million dollars a year on AI/ML and I don’t think anyone on the team can tell me if it’s working.” We are building software to fix that.
ML/AI Industry Complexity
There are platforms that are working to be the all encompassing end-to-end solutions to do Data Prep, Model Building and Production. We have experienced the inflexibility that lock-in creates with an end-to-end platform. As we talk with industry teams, we’ve built a deep belief that there is a huge need for best in breed vertically focused solutions.
We have a simple proposition in a complex industry. We are focused on one vertical, helping companies with their production ML and plan to do it immensely well.
Many software engineering best practices have direct analogs in model building. If you look at the software engineering space, your software solutions GitHub, IDE, production monitoring are not all the same end-to-end system. There are reasons why they are different pieces of software, they provide very different functions with clear differentiation.
Putting together the best best-of-breed products across the model pipeline makes a lot of sense for many teams today. Even if teams use end-to-end platforms, having a best-of-breed vertical solution that integrates on top of the end-to-end platform will add a lot of value.
We are at a nexus of more companies using more models, deployed in more critical business functions, than ever. There is a flood of investment in model building tools that is accelerating the number of models built annually by businesses. There is a real inflection point in the use of models in the real world.
The time is now.
Our team is incredibly passionate about making ML work well in live environments.
Arize has arisen.