开发者开源自托管法律AI平台 CaseCite，对标融资5亿美元的Harvey

开发者 Thomas Unise 发布并开源了名为 CaseCite 的法律研究平台，旨在作为商业产品 Harvey（已融资5亿美元）的替代方案 1。CaseCite 是一个支持自托管的解决方案，允许律师事务所上传文档并在本地基础设施上进行案件和法官研究 1。

该平台采用 Apache License 2.0 许可证，但 1.3.0 及之前版本适用 Elastic License 2.0 1。在数据存储策略上，文档、搜索索引和聊天记录均存储在用户自己的服务器上 1。为强调数据隐私，平台使用 BYOK（自带 API 密钥）模式，将用户数据仅发送至用户配置的 AI 提供商，而非平台维护者，且系统无遥测功能 1。

技术架构方面，CaseCite 基于 Docker + Docker Compose 构建，默认集成 OpenAI 进行嵌入和聊天，同时支持 Anthropic 或 Gemini 1。案例法来源主要依赖 CourtListener API 1。该项目版权归属于 Thomas Unise，标注年份为 2026 年 1。

Developer Thomas Unise has released and open-sourced CaseCite, a self-hosted legal research platform positioned as an alternative to the commercial product Harvey 1. The launch follows reports that Harvey raised $500 million in funding for its legal AI services 1.

CaseCite allows law firms to upload documents and conduct case and judge research on their own local infrastructure 1. To ensure data privacy, the platform stores all documents, search indexes, and chat records on the user's server rather than centralized cloud storage 1. It employs a Bring Your Own Key (BYOK) model, meaning user data is sent only to the AI provider configured by the user, not to CaseCite maintainers, and the software includes no telemetry features 1.

The platform is distributed under the Apache License 2.0, although versions 1.3.0 and earlier operate under the Elastic License 2.0 1. Technically, CaseCite is built using Docker and Docker Compose, with default integration for OpenAI embeddings and chat, while also supporting Anthropic or Gemini models 1. For case law data, the system primarily relies on the CourtListener API 1. The copyright for the project is held by Thomas Unise, dated 2026 1.