MoonshotAI
Featured
Kimi-K3
Kimi K3 is an open-weight, 2.8T-parameter native multimodal agentic model with a 1M-token context window, designed for frontier coding, knowledge work, and reasoning tasks.
AI repository intelligence
FR-AI / ANALYSISWhy it is worth attention
It ships as a standalone Go runtime that embeds FrankenPHP, Mercure, and a full Laravel app, making self-hosting as simple as downloading a binary or running a Docker container.
Who it is for
- Small and medium support teams seeking self-hosted AI support
- Developers and SMBs wanting control over customer data
- Multilingual teams needing automatic translation in their inbox
- System administrators evaluating on-premises customer service software
Use cases
- Managing all email, chat, and other omnichannel conversations in one AI-assisted inbox
- Using AI to draft replies, summarize conversations, and provide automated support
- Translating customer messages and previewing replies in the visitor’s language
- Building and testing knowledge bases for AI retrieval and support
Strengths
- Omnichannel unified inbox with contacts, tags, custom attributes, and quick replies
- AI-native features include automated support, reply assistance, summaries, and knowledge base retrieval testing
- Automatic translation keeps original and translated text together and offers pre-send visitor-facing previews
- Standalone Go runtime delivers simple Linux systemd, Windows, and Docker deployment with backup/restore commands
Considerations
- Automatic translation requires configuring a translation provider; no built-in translation service is described
- Self-hosting brings operational duties such as TLS setup, service management, storage, and backups
- Prebuilt distribution focuses on Linux and Windows plus Docker; local development requires PHP 8.5 ZTS, Go, Node.js, and other toolchains
README quick start
HelmDesk
English | 简体中文
HelmDesk is an open-source, self-hosted, AI-native customer support system for small and medium-sized teams. It brings omnichannel customer conversations into one workspace for AI-powered support and team collaboration.
Screenshot Preview
Unified Inbox
Dashboard
Contact Management
Knowledge Bases
Automatic Message Translation
With a translation provider configured, HelmDesk can translate incoming and outgoing messages into each teammate's preferred language. The inbox keeps the original text and translation together, preserving the conversation context when agents hide or refresh a translation.
Translate Before Sending
Agents can write a reply in the language they know best, preview the translation the visitor will receive, and confirm it before sending.
Key Features
- Omnichannel conversation intake and a unified inbox
- AI-powered automated support, reply assistance, and conversation summaries
- Automatic inbox translation and visitor-facing reply previews
- Knowledge base management and retrieval testing
- Contacts, tags, custom attributes, and quick replies
- Team members, roles, and permission management
- AI providers, models, object storage, and integration settings
Conversation Lifecycle
The diagram below shows initial routing, AI and human handoffs, timeout-driven transitions, closing, and reopening.
Technology Stack
- PHP, Laravel 13, Laravel Actions, Laravel Data
- Vue 3, Inertia.js 3, TypeScript, Tailwind CSS 4
- Go, FrankenPHP, Mercure
- SQLite, TNTSearch
Local Development
PHP 8.5 ZTS, Composer, Node.js, Go, and php-config are required.
git clone git@github.com:helmdesk-ai/helmdesk.git
cd helmdesk
composer setup
make
Once started, open http://localhost:8080.
Building
Build artifacts are stored in build/output.
Every build requires a SemVer such as 1.0.0. Make commands use APP_VERSION, and Po
Description
AI-native, self-hosted customer support for small and medium-sized teams. 面向中小团队的 AI 原生自托管客服系统。
Related repositories
Similar projects matched by category, topics, and programming language.
lopopolo
Featured
harness-engineering
Harness Engineering is a methodology for improving coding agent outputs by carefully crafting the environment around them—providing curated context, tools, and executable constraints that encode an organization’s nonfunctional requirements and cumulative lessons.
slvDev
Featured
esp32-ai
A 28.9 million parameter language model runs on an $8 ESP32-S3 microcontroller entirely on-device, generating simple stories at about 9.5 tokens per second.