Kite Raises $17M to Advance AI-Powered Code Completion
San Francisco startup Kite secures $17 million in funding to expand its machine learning-based code completion tool. The platform now allows developers to run its AI models locally without an internet connection.
San Francisco-based startup Kite raises $17 million in a new funding round led by Trinity Ventures to enhance its AI-driven code completion tool. The startup uses machine learning to predict what developers are trying to type, moving beyond the traditional alphabetical sorting used by standard autocomplete features. By analyzing repetitive coding patterns, Kite aims to save programmers the time they typically spend searching through documentation or debugging basic errors on sites like Stack Overflow.
Kite now offers significantly smarter capabilities by predicting multiple words ahead instead of just the next snippet of code. To achieve this, the team builds custom machine learning models specifically designed for programming languages, as standard natural language processing models fail to understand code structure. The system trains on all available Python code hosted on GitHub, allowing the correct suggestion to appear in the top three results 37 percent of the time.
The updated tool runs entirely locally on a developer's machine, removing the need for a constant internet connection. While the software currently only supports Python, it integrates seamlessly with popular editors like VS Code, Vim, and PyCharm, attracting roughly 30,000 active users. Ultimately, Kite's founder envisions a future of "fully automated programming" where developers simply tell computers what to do in high-level language, though he acknowledges this Star Trek-like reality requires a few more AI breakthroughs.