English | 中文

XpertAI Cloud · Self-hosting · Documentation · Enterprise inquiry

Open-Source AI Platform for Enterprise Data Analysis, Indicator Management and Agents Orchestration

Xpert AI is an open-source enterprise-level AI system that perfectly integrates two major platforms: agent orchestration and data analysis.

🚀 3.10 Agentic Apps & File Understanding!

XpertAI 3.10 turns plugins into first-class Agentic Apps, combining plugin-provided Xpert templates, Workbench views, runtime middleware capabilities, and file-aware agent execution into a more extensible assistant platform.

• Build business apps with plugin targetApps and targetAppMeta, plugin-contributed Xpert templates, deprecation controls, and richer manifest metadata for app-specific plugin centers.
• Render plugin-owned Workbench experiences with fixed views, remote component iframe entries, host theme propagation, host events, task status updates, and client commands.
• Open tool-triggered remote components in the Assistant Workbench, including the built-in Data X metric management plugin for project and semantic model metric operations.
• Use the new File Understanding layer to parse uploads into FileAsset records, artifacts, chunks, page images, citation anchors, and workspace paths that agents can search and read on demand.
• Give middleware safer runtime capabilities for files, knowledgebases, assistant tasks, organization context, chat events, and the new Office Automation middleware flow.
• Monitor Copilot operations with user-level usage details, usage summaries, filters, checkpoint retention cleanup, Prometheus metrics, and conversation goals.

In today’s rapidly evolving AI landscape, enterprises face a key challenge: How to balance the creativity of LLMs with the stability of workflows? Pure agent architectures are flexible but hard to control; traditional workflows are reliable but lack adaptability. Xpert AI’s Agent-Workflow Hybrid Architecture is designed to resolve this conflict, enabling AI to have “free will” while adhering to “rule-based order.”

Blog - Agent-Workflow Hybrid Architecture

By coordinating the collaboration of multiple intelligent agents, Xpert can handle complex tasks. Xpert integrates different types of AI agents through efficient management mechanisms, leveraging their capabilities to address multidimensional problems.

A cloud-based agile data analysis platform supporting multidimensional modeling, metrics management, and BI visualization. The platform connects to various data sources, enabling efficient and flexible data analysis and visualization, and offers multiple intelligent analysis tools to help enterprises quickly and accurately uncover business value and make operational decisions.

Before installing Xpert, make sure your machine meets the following minimum system requirements:

• CPU >= 2 Core
• RAM >= 4 GiB
• Node.js (ESM and CommonJS) - 20.x, 22.x

The easiest way to start the Xpert server is through docker compose. Before running Xpert with the following commands, make sure that Docker and Docker Compose are installed on your machine:

cd xpert
cd docker
cp env.example .env
docker compose up -d

After running, you can access the Xpert dashboard in your browser at http://localhost/onboarding and start the initialization process.

Please check our Wiki - Development to get started quickly.

Xpert AI Platform Demo at https://app.xpertai.cn.

Notes:

• You can generate samples data in the home dashbaord page.

Xpert AI Platform SaaS is available at https://app.xpertai.cn.

Note: it's currently in Alpha version / in testing mode, please use it with caution!

• TypeScript language
• NodeJs / NestJs
• Nx
• Angular
• RxJS
• TypeORM
• Langchain
• ECharts
• Java
• Mondrian

For Production, we recommend:

• PostgreSQL
• PM2

• SDK – streamlines access api of the XpertAI platform.
  • SDK (TypeScript)
    • digital experts
    • long-term memory storage
    • contextual files
    • Conversation management
    • knowledge bases
  • SDK (Python)
• Plugins – extensible plugin system.
  • Plugins system
  • Marketplace to showcase the plugins ecosystem.
  • Hot-swappable plugin system.
• Chatkit – front-end component library for embedding digital expert chat dialog.
  • ChatKit javascript version.
• Widgets – UI widgets that let large-model responses drive richer interface experiences.
• Agent Middlewares Plugin-based Agent Middleware.
• Agent Skills – lightweight agent skills for rapid custom capability integration, offering a quicker alternative to MCP tools.
• Project Management – AI-driven project execution workspace for planning, coordinating, and running multi-agent work.
  • Project workspaces with overview, Kanban board, files, teams, and tabbed project layout.
  • Project, sprint, task, backlog, and swimlane models with APIs for planning and execution strategy.
  • Bind published Xperts as project assistants and project teams, with team role and environment settings.
  • Project assistant tools for backlog management, sprint planning, task updates, team binding, and runnable task dispatch.
  • Task assistant ChatKit conversations with execution artifacts and project-context handoff.
  • Project event stream for real-time board updates and agent execution feedback.
• Audit, Security, Compliance – Enterprise-grade features to ensure data privacy and compliance.
  • Audit logs
  • Role-based access control
  • Data encryption
• Sandbox – Secure testing environment, isolating experiments from production.
• Trace, Evaluation – Observability and evaluation tools for agents and workflows.
  • Trace system
  • Evaluation framework
• System monitoring and alerting
  • Sentry integration
  • Prometheus integration

• For business inquiries: mailto:service@xpertai.cn
• Xpert AI Platform @ Twitter

We support the open-source community.

This software is available under the following licenses:

• Xpert AI Platform Community Edition
• Xpert AI Platform Enterprise Edition
• Xpert AI Platform Enterprise Pro Edition

Contributors

• Please give us ⭐ on Github, it helps!
• You are more than welcome to submit feature requests in the Xpert AI repo
• Pull requests are always welcome! Please base pull requests against the develop branch and follow the contributing guide.