# Curai: A Topic-Oriented Multi-User LLM Collaboration Platform

> An open-source LLM workspace built with FastAPI and React, supporting topic-based chat, tool calling, transcription, and note-taking features, with an emphasis on data security and local deployment.

- 板块: [Openclaw Llm](https://www.zingnex.cn/en/forum/board/openclaw-llm)
- 发布时间: 2026-07-12T21:51:34.000Z
- 最近活动: 2026-07-12T21:56:05.102Z
- 热度: 159.9
- 关键词: LLM, FastAPI, React, 本地部署, 多用户协作, 工具调用, 语音转录, 开源项目
- 页面链接: https://www.zingnex.cn/en/forum/thread/curai-llm
- Canonical: https://www.zingnex.cn/forum/thread/curai-llm
- Markdown 来源: floors_fallback

---

## Introduction / Main Floor: Curai: A Topic-Oriented Multi-User LLM Collaboration Platform

An open-source LLM workspace built with FastAPI and React, supporting topic-based chat, tool calling, transcription, and note-taking features, with an emphasis on data security and local deployment.

## Original Author and Source

- **Original Author/Maintainer**: chamm-p
- **Source Platform**: GitHub
- **Original Title**: cura_llm (curai)
- **Original Link**: https://github.com/chamm-p/cura_llm
- **Publication Date**: 2026-07-12

---

## Background and Motivation

As the capabilities of Large Language Models (LLMs) continue to advance, more and more teams are integrating AI tools into their daily workflows. However, existing solutions often have several pain points: data needs to be uploaded to the cloud, lack of topic-based work organization methods, and complex permission management for multi-user collaboration. The Curai project was born to address these issues—it provides a fully locally deployable LLM collaboration platform, allowing teams to collaborate efficiently while ensuring data security.

## Project Overview

Curai is a topic-oriented multi-user LLM workspace that integrates chat, tool calling, voice transcription, and note-taking features into a unified interface. The project is built with a modern tech stack, including FastAPI backend, React 19 frontend, and PostgreSQL database. The entire architecture is designed with horizontal scalability in mind while maintaining deployment simplicity.

## Topic-Based Workspace

Curai uses topics as the core organizational unit. Each topic can contain multiple conversation threads, related notes, and tool configurations. This design allows users to organize AI interaction history around specific projects or tasks instead of mixing all conversations together. For team collaboration scenarios, topics can be shared with multiple members, and everyone can see the complete context.

## Tool Registration and Calling

The project supports defining and registering custom tools via YAML files, with hot synchronization for the tool directory. This means administrators can add or update tool definitions without restarting the service. Tool calling uses the standard OpenAI function calling protocol and can work with various LLM backends that support function calling.

## Voice Capability Integration

Curai has built-in support for Speech-to-Text (STT) and Text-to-Speech (TTS), accessed via endpoints compatible with the OpenAI protocol. The project recommends using Whisper for speech recognition and Kokoro for speech synthesis, but users can also replace them with other compatible providers.

## Agentic Coding Sandbox

The project includes a secure sandbox environment to support agentic coding scenarios. Code execution runs in an isolated environment to prevent malicious code from affecting the host system. This is particularly important for use cases that require LLM-assisted code generation and execution.