TruEra, a startup that offers an AI quality management solution to optimize, explain and monitor machine learning models, today announced that it has raised a $25 million Series B round led by Menlo Ventures. Existing investors Greylock Partners, Wing Venture Capital (which led its $12 million Series A round in late 2020), Harpoon Ventures, Conversion Capital, the Data Community Fund, as well as new investors Forgepoint Capital and the B Capital Group’s Ascent Fund also participated in this round. In total, TruEra has now raised $42.3 million.
“We believe that the next big challenge in AI is the quality challenge,” TruEra CEO and co-founder Will Uppington said. “AI is at an inflection point: lots of opportunities but also a lot of challenges to make AI actually work in the enterprise. And we think that’s the major issue that’s preventing AI systems from getting into real-world use and actually delivering on KPIs.”
Image Credits: TruEra
Uppington argues that it’s not just hard to design and build high-quality models to begin with, but there’s also still a lot of concerns around trust, transparency and fairness when it comes to putting models into production — and increasing regulatory pressure around AI fairness is giving enterprises pause because they need to put auditable systems in place to conform to these rules. And once a model can finally be put into production, businesses have to ensure that quality remains high, even as some of the underlying data changes.
TruEra argues that an enterprise AI quality management solution needs to approach these problems head-on, starting with tools that developers can use while they train the model so that they can test and evaluate their models long before they go into production. To do this, the company’s service can be integrated right into the kind of Jupyter notebooks that most data scientists are already using to build their models anyway, for example.
“We’re in the space where software development was in the 90s, before you had tools and agile development methodologies,” Uppington said. “Data science is very waterfall and the models are still pretty black box. That reduces the quality of your development process, just like it did in software development in the 90s. […] We think that we can help the world get to those better tools and more agile-like development with this kind of comprehensive, fast quality testing and making it really easy for the data scientists to use it.”
The company says it saw its revenue grow over 5x since it raised its Series A round in late 2020. In part, Uppington noted, that’s driven by the fact that a lot of enterprises are now getting to the point in their AI journey where they want to put models into production and are starting to face these quality challenges. Add to that the regulatory environment and some high-profile failures (think Zillow), and it’s a good time to be in the AI quality space.
Menlo Ventures partner Tim Tully, the former CTO of Splunk, also stressed that TruEra approaches the problem through the lens of the model, with co-founder and Chief Scientist Anupam Datta having done some of the early academic work on AI explainability as a professor at Carnegie Mellon.
“If I were the machine learning data scientist, what would I want to use? I looked at all the companies and I was looking for the one that provided the depth and approached the problem through the lens of the model, as opposed to just the opposite direction, which I think is completely wrong,” Tully said. “And frankly, I think a lot of the products are very superficial in the treatment of the problem. I want to go really deep with it and I want to see some proprietary research that creates distance from the competitors.”
Taking open source Python libraries and putting a user interface on top of them doesn’t quite cut it in this market, Tully added.
