A year and a half into Codeium’s existence, and we’ve made some pretty big strides to become a leader in the code assistant space. We have over 500k active users and over 500 paying enterprise customers, with growth not showing signs of slowing down, and that has led to some pretty humbling recognitions…
Forbes AI 50
Codeium was included in the 2024 Forbes AI 50 list, a list of the top privately held companies in AI, and was the only AI code assistant to make the list. While we appreciate the recognition, we are working even harder to continue to cement Codeium as the tool every developer uses to dream bigger.
JetBrains Command
Codeium Command, our popular in-editor instructive modality in Visual Studio Code, has now come to JetBrains. We have trained the fastest and most powerful instruction following model for code in order to power Codeium Command, making it an incredibly snappy and powerful experience.
Custom context
In Visual Studio Code, custom context allows you to specify natural language that will be added to every request that goes to Codeium, either in autocomplete or chat. For example, if you are using a particular framework (e.g. Tailwind CSS), you can just specify this in custom context (e.g. “Use Tailwind CSS”), and Codeium will know to generate all code with that specification. Custom context is a very simple way for you to guide the AI to do what you want.
Indexing access controls
We have added support for Okta as a source for defining these IAM groups, which now control both indexing access and sub-team analytics.
Pin remote sub-repo context items
We used to only have support for pinning entire remote repositories, but if these remote repositories are very large, then this pinning does not achieve much in guiding the AI to the relevant files and directories. Now, you have the ability to pin sub-repo entities such as directories, files, and individual context items (such as functions and classes) in the context pinning feature. Another easy way to help the AI to help you!
VertexAI Gemini integration
As we often say, Codeium is not a model company. While we have trained our own proprietary code-biased LLMs to be fast, really good at coding tasks, and using existing context, sometimes we do want a general purpose reasoning LLM to answer more generic questions. While we already support OpenAI integrations via a GPT-4 option in Codeium Chat for the Teams tier, we want to increase optionality for these large generic models that you may already have available.
Content
Some content on Codeium from around the Internet:
- Demo Video: A single video to see in action almost all features of Codeium to date so that you know how to best use the tool.
- World Wide Technology Case Study: WWT has seen 30-50% productivity improvements on boilerplate tasks internally, and we are excited to work together to bring the same wins to more joint customers.
- Hunters & Unicorns: Our VP of Sales, Graham Moreno, appeared on the Hunters & Unicorns podcast to talk about Codeium and why he joined.
- Google Gemini CodeAssist Review: As we always do, we try to review every major new tool on the market.
As always, thank you for being part of the journey. We haven’t mentioned it recently, but if you have not joined our Discord yet, we strongly recommend doing so!