Refact.ai is an open-source AI coding assistant that boosts developer's productivity. It's suitable for both individual and enterprise use, supporting 25+ programming languages including Python, Javascript, Java, Rust, Typescript, PHP, Java, C++, C#, Go and more.
Simplify your coding experience with Refact.ai's functions:
- Code-completion: helps you write code faster. It looks up and down from the cursor position in the current file, as well as other files in your project for context. After fine-tuning (available in Self-Hosting and Enterprise versions) a code completion model will provide you with more relevant suggestions: it can memorize your coding style, the right way to use your internal APIs, and the tech stack you use.
- AI Toolbox: helps explain, refactor your code, write documentation, find bugs, you can ask in natural language to generate new code for you. The Toolbox is customizable, you can create your own commands.
- Integrated in-IDE Chat: It's better than switching between IDE and a browser: select a piece of code in the editor, ask a question without the need to copy-paste anything, and apply changes back as a diff. Chat can either automatically include the relevant files within the project for context, or you can use @-commands to control what context to use to answer your question.
- Self-hosting or cloud version: enables you to host code completion and chat models yourself. The Refact.ai self-hosting server supports many recent models, offers efficient inference, and facilitates fine-tuning.
Are you part of an Enterprise and want to boost developers' efficiency by 45%? The Refact.ai Enterprise plan offers:
- Knows Your Code: Refact.ai focuses on automated customization for internal codebases and improving over time.
- Ultimate Security: Deploy Refact.ai on-premise: on your own servers or private cloud. Your data never leaves your control. Telemetry from the plugins goes to your server and nowhere else. You can verify what the code does, it's open source.
- Fine-Tuning: We've solved all the complexity for you, no data engineer required. A fine-tuned code completion model will provide you with more relevant suggestions: it can memorize your coding style, the right way to use your internal APIs, and the tech stack you use. After fine-tuning, auto-completion helps our clients to write around 45% of their code.
- User Statistics: Keep track of how much code Refact.ai helps to write for each team member. Measure productivity gains with real numbers.
- Efficient Inference: The Enterprise version supports paged attention and continuous batching, supporting up to 50 engineers per GPU for code completion.
- Priority Support: Our engineers are always available to assist you at every stage, from setup to fine-tuning and beyond.
- Access to 20+ LLMs: (including Refact/1.6B, GPT-4, Code Llama, StarCoder2, Mistral, Mixtral, etc.) some of them with the opportunity to fine-tune.
To get a 2-Week free trial for your team, just fill out the form on our website. We'll reach out with more details!
Connect with other developers who care about AI tools in our Discord community. Ask questions, share your opinion, propose new features.
Implement a feature you want, and contribute to the future of programming!