Today DeepCode Swiss startup that uses artificial intelligence and machine learning to automate code analysis, announced the receipt of investments in the amount of $ 4 million from Earlybird venture capital funds, 3VC and Btov Partners. The company plans to spend these funds for implementation support in their service of new programming languages, and product marketing in the global it market.
The analysis of the code required to detect errors, potential security vulnerabilities, violations of generally accepted formatting style, and much more in the early stages of software development before the code is anywhere used. Usually this procedure is performed in parallel with the development of new code and immediately after it is completed, anticipating the moment of testing. “Software testing looks at code from the outside, but code analysis allows you to look at it from the inside, “explains co – founder and CEO DeepCode Paskalev Boris (Boris Paskalev) in an interview with VentureBeat.
Most often, the checking code is performed by the authors together with colleagues and managers to identify obvious mistakes before moving on to the next stages of development. And the larger the project, the more lines of code you need to verify which takes a significant amount of time programmers. Tools that should speed up the process, there are already quite a long time, for example, static code analyzers such as Coverity and PVS-Studio, but they are usually limited in their abilities, focused on “annoying and repetitive stylistic problems, formatting and small logic errors,” explains Paskalev.
DeepCode, in turn, covers a wider range of problems, for example, detecting such vulnerabilities as opportunities for cross-site scripting and SQL injection, as its built-in algorithms not just analyze the code as a set of characters, but trying to understand the meaning and purpose of the written program. This is based on a system of machine learning which uses training billions of lines of code from publicly available open source code. DeepCode analyzes previous versions of the code and subsequent changes that were made to it, to examine what errors and how to correct the real programmers in their work, and then offer similar solutions for their users. In addition, the system uses traditional forecasting algorithms to search for potential problems in your code like the above mentioned static analyzers.
One of the key issues when using DeepCode: how reliable is the automatic code review? The accuracy of the analysis is less than 100% means that developers still have to analyze your code manually. In that case, how much time will actually liberate the use of tools to automate this task? According to Paskaleva, DeepCode can save developers approximately 50 % of the time they currently spend finding errors yourself, which is quite a significant figure.
Developers can plug DeepCode to their accounts on GitHub or Bitbucket, and the tool supports local configuration Networks. Additionally, the project has a special API that allows developers to integrate DeepCode the private system for the development. After connecting to the repository DeepCode will analyze each change of the code and to note potential problems.
“On average, developers spend about 30 % of their time on search and elimination of mistakes, but DeepCode can save half of that time now and in the future even more“, – says Boris. “Because DeepCode learning directly from the global community of developers, it is able to detect more problems than it could ever find one person or a group of reviewers“.
In addition to today’s news about getting investment DeepCode also announced new security policy of your product. Still DeepCode was free only for projects on software development with open source. Now he will be free to use for any educational purposes and even for commercial companies with less than 30 developers. It is obvious that with this step the creators DeepCode want to make your product more popular among small teams. Additionally DeepCode charges $ 20 with the developer to deploy solutions in the cloud and 50 dollars from a local developer for support.