$10M worth Mabl brings machine learning to software testing

Mabl, a startup that’s coming out of stealth today, uses machine learning to make functional testing for developers as easy as possible. Mabl users don’t have to write extensive (and often brittle) tests by hand. Instead, they show the application the workflow they want to test and the service performs those tests — and even automatically adapts to small user interface changes.

The Boston-based company also today announced that it has raised a $10 million Series A funding round from CRV and Amplify Partners.

It’s worth noting that the team behind Mabl has a bit of a pedigree. Co-founders Izzy Azeri and Dan Belcher previously founded Stackdriver, the cloud monitoring solution Google acquired in 2014. After a few years at Google, during which time the company deeply integrated Stackdriver into its Cloud Platform, the founders decided to try something new, though, and left Google last January. “As we met with hundreds of software teams, we latched on to this idea that developing — the process of writing new code and integrating it with your code bases — is very fast now, but there’s a bottleneck in QA,” Azeri said. “Every time you make a change to your product, you have to test this change or build test automation.”

So with Mabl, the team is trying to bring this process into the 21st century by combining functional testing (that is, actually trying out the front end of the application and looking at the results) with machine learning. Right after signing up for Mabl, the service starts scanning your site and starts looking for potential errors. You then create your own tests by walking Mabl through a scenario (with the help of a Chrome plugin) and the service automatically understands that a given button or link leads to a certain action. As long as the service can somehow identify that button, for example, it’ll perform the same test, even if you are experimenting with a new user interface and move it to a new position on your page. “If there’s something unique, we’ll find it, even if it’s in a different spot,” explained Belcher.

Mabl will alert developers of any visual changes, JavaScript errors, broken links and increased load times.

Given that the founding team came from Google, it doesn’t come as a surprise that Mabl is running on the Google Cloud Platform. But as Azeri and Belcher told me, they actually gave their engineers free rein to pick AWS or another platform, too. Google won that vote and the team now makes heavy use of tools like the Google Kubernetes Engine for spinning up containers for tests, Google Cloud Functions, BigQuery, Cloud ML Engine and App Engine.

During the preview period, the service will be free to all developers. After that, the team — which currently consists of about 20 people — will start monetizing the service, likely with a focus on charging users for the amount of testing they do. What exactly this will look like, though, still remains to be seen.

Current Mabl users, who started testing the service during its closed beta, include the likes of RunKeeper, creative agency 24G and Codeship. “Mabl frees up Codeship’s team to build product, and that’s great for our customers. Nobody on the team has to worry about testing; everyone knows if there’s an issue, Mabl will tell us, then we’ll fix it and move on,” said Moritz Plassnig, the co-founder and CEO of Codeship.

As Azeri and Belcher noted, funding this new company was a bit easier than raising funding for Stackdriver. CRV, they told me, had wanted to invest in Stackdriver but missed out on that opportunity (and the company’s exit to Google), so it’s maybe no surprise that CRV partner Murat Bicer really wanted to invest in the founders this time around. Indeed, Bicer is joining Mabl’s board of directors.