Anomalo launches with $33M Series A to automatically find issues in data sets

As companies gather ever-growing sets of data, finding issues with that data that could impact the viability of a machine learning model becomes increasingly important. Anomalo is putting machine learning to work to help solve the data viability issue automatically.

Today the company announced a $33 million Series A investment led by Norwest Venture Partners with participation from Two Sigma Ventures, Foundation Capital, First Round Capital and Village Global.

The company was founded by two Instacart veterans who worked at solving similar problems at their previous company. Elliot Shmukler, co-founder and CEO at Anomalo, said that if you’re counting on data to run your business, any issues in that data can be problematic for the organization.

“What Anomalo does is it connects to these enterprise data warehouses like Snowflake, where they’re stockpiling all of this data that companies collect, and it monitors all those data sets for unusual issues and unwelcome changes in that data, which can cause lots of issues if you’re actually trying to rely on that data to run your business,” Shmukler explained.

It sounds simple enough, but what Anomalo is doing behind the scenes is connecting to these data warehouses and training a machine learning model on what is normal for this particular set of data and reporting when it finds issues. Shmukler says that this approach is in contrast to other solutions, which force data teams to explicitly define what good data looks like, a method he says becomes increasingly unmanageable as the number and size of the data sets grow.

“If you look at other solutions … they require folks on the data team to go in and essentially define the expectations for them, to say, this is what good data looks like, [and] that’s a tremendous amount of work. As your data changes and you launch new products and new geographies, you have to keep updating those definitions,” he said.

It was a problem the founders saw when they were on the data team at Instacart, where they had to constantly update these definitions. When they launched Anomolo, one of their goals was to automate that process for data teams so they didn’t have to deal with that manual work.

It wasn’t an easy problem to solve. The two founders — Shmukler and CTO Jeremy Stanley — left Instacart in 2018 to launch the company and it took a couple of years to get that machine learning model to work the way they wanted it to, without too many false positives or requiring too much history as a basis for learning.

While the founders didn’t want to reveal the exact number of current employees, the plan is to hire another 40 or 50 in the next year. Shmukler says that when he and Stanley decided to start a company, they set core values that included diversity.

“We actually wrote down a set of values for the organization to abide by, and one of them was being diverse. That was something very important to us at Instacart and something that we just wanted to continue working on [at this company]. And so we’re very mindful of making sure that when we’re recruiting for a role that we bring in a diverse set of candidates for that role … and the good news is that it’s working, at least today, where 25% of our engineering team are women, which is probably unusual for an early-stage company. And we hope to keep that going and continue to improve that,” he said.

While the company is formally launching today, it has paying customers and reports that it has at least $1 million in revenue already. It charges by the data set it’s monitoring rather than by the user or data coming through its pipeline. Customers out of the gate include BuzzFeed, Discover Financial Services and Substack.