Codeium Inc., a startup with a developer productivity platform powered by large language models, today announced that it has secured a $65 million Series B funding round.

Kleiner Perkins led the investment with participation from General Catalyst and returning backer Greenoaks. According to Bloomberg, the raise values Codeium at $500 million. The company will use the capital to grow its sales team and build more features for its developer productivity platform. 

“Software engineering is the backbone of innovation yet the process is expensive and inefficient due to imperfect knowledge retrieval, boilerplate tasks, maintenance overhead, and more,” said Codeium Chief Executive Varun Mohan. “The fast rise of LLMs is now giving developers the opportunity to overcome these challenges and significantly increase their productivity.”

The platform takes the form of an extension that software teams can install in their IDEs, or integrated development environments. An IDE is a kind of word processor geared toward writing code rather than prose. Such applications usually also provide a range of other features, notably tools for finding and fixing bugs.

After it’s installed, Codeium embeds an artificial intelligence chatbot into the user’s IDE. It appears as a new sidebar on the left side of the interface. Developers can ask the artificial intelligence for information about the application they’re working on, such as how two unfamiliar code components interact with each other.

Codeium says its chatbot saves time in other ways as well. According to the company, a developer can input a natural language description of a task and have the software automatically generate the code necessary to perform that task. It can generate code in more than 70 programming languages, as well as translate between them.

Alongside the embedded chatbot, the company’s platform provides an autocomplete tool. It connects to the section of an IDE’s interface in which developers write code and suggests additions directly within the panel. The tool can generate up to several lines of code at a time.

According to Codeium, rival products struggle to generate useful autocomplete suggestions when they’re asked to make an addition between two existing code snippets. The company claims its platform provides the only autocomplete tool on the market that doesn’t have this limitation. As a result, developers can apply it to a broader range of programming tasks. 

Under the hood, the platform is powered by a set of internally-developed large language models and OpenAI’s GPT-4. The company says that its models can generate not only code but also text. That allows them to generate documentation explaining how a newly written piece of code works.

Using the $65 million round announced today, Codeium plans to extend its platform’s AI capabilities to additional areas. One feature currently in the works will ease the task of fixing cybersecurity vulnerabilities and other bugs. It also intends to introduce an AI tool for modernizing applications’ legacy code components.

The company provides its platform in a free edition and two paid versions that offer additional features, notably administrative controls geared toward enterprises. Since releasing the platform in late 2022, the company has built up an installed base of more than 300,000 developers. Codeium says the software is also used by several Fortune 500 companies.

Image: Codeium