Because AI technologies provide the benefits of faster learning and improved productivity, developers are prepared to use them. In a recent poll, 77% of developers felt optimistic about incorporating AI into their workflows, and 70% said they plan to utilize AI coding tools this year.
Additionally, investors see promise in generative coding tools, especially in what they can do on an enterprise level. And for businesses like Tabnine, which recently disclosed that it raised $25 million in a Series B fundraising round with participation from Atlassian Ventures, Elaia, Headline, Hetz Ventures, Khosla Ventures, and TPY Capital, this enthusiasm is translating into additional money.
To establish a platform that incorporates generative AI into different stages of the software development lifecycle, Dror Weiss and Eran Yahav co-founded Tabnine in 2012. Weiss is a graduate of the Israel Institute of Technology’s Technion, where Yahav was, and a computer science professor.
Tabnine offers Tabnine Chat, an AI “code assistant” that creates code and responds to inquiries about an organization’s codebases, similar to ChatGPT for code, and additional coding tools driven by first- and third-party generative AI models.
Amazon CodeWhisperer and GitHub Copilot are two rivals of Tabnine. However, Weiss claims that compared to competing systems, the firm offers greater customisation and control. For instance, clients can use a virtual private cloud or on-premises to implement the tools.
In an email conversation with TechCrunch, Weiss said, “Our flexible architecture means we can switch [code-generating AI] models relatively easily and are thus not ever competing with the big generative AI model builders.” “We can provide developers with access to new models from other vendors and adapt with AI as it develops; Tabnine can enable developers to use those models wherever they code.”
Weiss also argues that, from a business standpoint at least, Tabnine poses less legal danger than its rivals.
In a class action lawsuit, Microsoft, GitHub, and OpenAI are being sued for allegedly breaking intellectual property laws by allowing Copilot, who was trained on billions of examples of publicly available code from the internet, some of which were subject to restricted licenses, to repeat portions of copyrighted code without giving proper credit. Some lawyers say businesses could be in legal trouble if they accidentally add copyrighted suggestions from AI systems like Copilot to their production software. This is in addition to the liability issue.
Weiss points out that Tabnine only employs AI models trained on code with permissive licenses or collaborates with clients to train models on their codebases.
“We have much better control and security because we use a curated data set and know what has gone into it,” Weiss stated. Additionally, thanks to this basis, our clients use private models built on their code and operating in their data centers and virtual private clouds.
Tabnine’s strategy is effective, especially considering the demise of Kite, one of its competitors, towards the end of last year. While not quite as large as Copilot’s, with around one million paying users and 37,000 corporate clients, Tabnine’s user base of over a million and 10,000 customers is still significant.
According to Weiss, Tabnine plans to use the cash from Series B, which increased the company’s total funding to $55 million, to enhance its generative coding skills and bolster its worldwide support and sales teams. Compared to now, Tabnine anticipates having 150 workers by the end of the year.
