Arthur raises $15M in a Series A round to fill the last-mile gap in the AI software stack

News provided by

Dec 09, 2020, 11:41 ET

NEW YORK, Dec. 9, 2020 /PRNewswire/ -- Arthur today announces that it has closed a $15M Series A funding round to scale operations and meet surging customer demand. The round was led by Index Ventures, and includes new investors Acrew and Plexo Capital. The investment will be used to accelerate delivery of key capabilities, making sophisticated AI monitoring, fairness tools, and model performance management available to all organizations.

Arthur's funding announcement signals a growing trend: companies need tools to provide oversight and control over their machine learning systems. Public incidents exposing the dangers of AI bias, including issues with OpenAI's GPT-3 and face detection failures from Twitter and Zoom, have made it clear that continuous bias monitoring and mitigation are essential for any organization looking to deploy AI. The pandemic has sent many organizations' models into a tailspin, due to unexpected behaviors like concept or data drift. They now know—some having learned the hard way—that they can't scale their AI operations without the right controls in place.

This growing need has driven 300% revenue growth at Arthur in the last six months from both enterprise customers like Humana and AI-first companies like Truebill. The Arthur platform gives their data science teams the same level of operational awareness found in domains like cybersecurity and application performance management.

"When you take your models out of the lab and put them in the real world, things can get unpredictable. The world is a dynamic place. AI is embedded in so many industries where we need to know how these models are working to ensure both high performance and transparency," says Co-Founder and CEO of Arthur, Adam Wenchel.

"Most data scientists would prefer to focus on solving the next big problem rather than babysitting their models. But nobody wants a runaway mistake that hurts their customers and their ROI. If we create better guardrails, teams can move faster and deploy more quickly without being reckless. We believe the future of AI is fairness and better bottom lines," says Mike Volpi, partner at Index Ventures who led the round for Arthur.

Founded in 2018, Arthur was built by a founding team with a combined 50+ years of enterprise AI experience. Arthur raised $3.3M in seed funding from Index, Homebrew, Work-Bench Capital, and AME Ventures in 2019, and all of its existing investors participated in this latest round. In addition to Wenchel, the Arthur co-founders include Liz O'Sullivan, John Dickerson, and Keegan Hines. Arthur became the platform they wished they had in previous roles: to provide visibility into the large-scale systems they built. They understood the importance of getting alerted to issues before small inaccuracies become million dollar problems. The team has seen firsthand that algorithmic bias and data drift are unavoidable, but that AI observability tools can help teams detect and combat these issues, before their models start behaving badly.

"Our team has many models in production. With the Arthur platform, checking their output and status at a glance is easy. We can spot issues early and with the ability to introspect the model run, easily diagnose and fix models reducing downtime. Couple that with custom alerting, we know there won't be any surprises with our ML operations. Arthur gives us confidence that our ML systems continue to achieve high rates of accuracy," says customer Chris Poirer, VP, Data at Truebill.

Businesses are recognizing that they aren't realizing the full potential of their models without a monitoring system in place. Arthur brings AI observability to organizations as they shift their focus from understanding how to build and deploy AI systems to understanding how to manage those systems in production—and to catch and prevent costly and harmful mistakes.

About Arthur:

Arthur is a proactive model monitoring solution that gives you the confidence that your AI deployments are performing as expected, and the peace of mind that you can catch and fix issues before they derail your models. No matter where your models are deployed, Arthur brings them all into view in a centralized platform. With advanced performance monitoring, bias detection, and customizable alerts, you'll never miss an issue; and Arthur's explainability engine makes debugging a breeze with prediction-level explanations, even for black box models.
https://www.arthur.ai

SOURCE Arthur