DeepSource, a member of the Winter 2020 Y Combinator cohort, announced a $2.6 million seed investment today. The company is building a solution to help developers automate static code analysis to find certain errors before going through human code review.
645 Ventures led the round with help from Y Combinator, FundersClub, Pioneer Fund, Liquid 2 Ventures and a slew of individual investors. The company had previously raised $140,000 in pre-seed capital.
DeepSource is taking advantage of a process called static analysis. Company co-founders Jai Pradeesh and Sanket Saurav are software engineers, and saw the static analysis tools that ship with language packages like Python and Go as overly complex and hard to use. Pradeesh said that as a result, most developers have not embraced them.
“What we’re trying to do is use static analysis to figure out if we can automate the objective parts of the code review. […] Whenever a software developer makes a commit or introduces any change in code, DeepSource automatically runs analysis on that code. It then flags issues like security issues, anti-patterns, progress performance issues and things like that,” Pradeesh explained.
By automating parts of the review process, it allows the human code reviewers to concentrate on the less objective parts of the review and find a myriad of issues before it even goes to human review. DeepSource gives the developer a report with the error and an explanation of how to fix it.
But the founders recognized the drudgery of constantly repairing the same sorts of errors over and over. So they created an automated repair tool called Autofix that goes through and fixes a set of common errors automatically for the developer.
The company currently has 12 people, including the two co-founders, based in Bangalore in India, but has plans to eventually move to the U.S. when the pandemic allows. It also plans to hire another 10-15 people over the next 12 to 18 months he said, primarily in engineering. The startup is working with YC advisors on building a diversity plan now, understanding that it is at a pivotal point in its evolution.
He says that the YC experience taught him and his co-founder about building more refined user personas for design and marketing purposes. Instead of broadly looking at the market as all developers, they could begin to focus on different aspects of that pool like a large team versus an engineering manager. He said they also learned to explore usage data to understand things like what kinds of errors were most common, which helped inform the creation of the Autofix feature.
Being part of YC also helped them when they had to leave the office and go remote earlier this year. Pradeesh admitted he had no experience running a remote team and his YC advisors helped him get comfortable with the process to the point they have good systems in place now.
“We put in these separate processes so that we do these bi-weekly sprints, and have learned ways to make sure everybody in the team communicates. Now that we have started placing these processes in place, the team is getting more and more used to it,” he said.