Refact.ai
Configurable and self-hostable AI code assistant for completion and chat, with fine-tuning on your own codebase
Overview
Refact.ai is a code assistant distinguished by its architectural flexibility: you can use it as managed cloud SaaS, self-host it on-premise on your infrastructure, or even fine-tune it on your own proprietary codebase for domain-specific completion and chat. The free plan includes 5,000 coins (internal billing unit) and full access to base completion and chat. The Pro plan ($10/month) unlocks unlimited coins, an expanded context window and priority access to premium models. One-time coin purchases are available (starting at $5 for 5,000 coins).
Enterprise on request adds complete on-premise self-hosting, advanced fine-tuning on proprietary codebase to improve accuracy, and dedicated support with guaranteed SLA. Refact exposes a publicly documented REST API and supports VS Code, JetBrains (IntelliJ, PyCharm, WebStorm), Neovim and text terminals. The interface remains English only, with no official French localization announced. The major absolute strength is total flexibility of having a fully configurable assistant with possible fine-tuning on your proprietary codebase to improve completion accuracy. The weakness is that this functional richness means increased complexity for technical setup, model configuration and long-term DevOps maintenance.
Our verdict
Best for development teams wanting a flexible code assistant, with option to fine-tune on proprietary codebase and ability to self-host for sensitive data. Not for you if you are seeking a plug-and-play solution without configuration: Refact's functional richness means increased complexity to set up and maintain.