Skip to content

Repository files navigation

Open Insight

๐ŸŒ Languages: English | ็ฎ€ไฝ“ไธญๆ–‡

Build the next generation of AI employees for your enterprise.

A team of AI employees designed to understand your business, perform specialized work, and deliver real business outcomes.

It enables AI to understand enterprise systems, enterprise data, enterprise knowledge and personal memory, then execute real business tasks through reusable Enterprise Skills.

Component Role License
ARP The runtime where AI employees work โ€” any model, any tool, any skill Apache-2.0
ONE-PI The reasoning engine โ€” virtual experts that understand, reason, and execute Apache-2.0
DMP The enterprise core โ€” four learning engines that teach AI your data, systems, and knowledge Commercial

Employees work in the browser, zero install. Deploy on your own infrastructure.

๐Ÿ“˜ Deployment Guide โ†’ ยท Full story: Organization Home


Architecture

Enterprise AI starts with understanding the enterprise.

Open Insight Architecture

Native AI Coding Architecture

ONE-PI connects to an extensible set of expert agents, each equipped with Prompt, MCP, API, and Skill capabilities.

ONE-PI Agent Architecture

Product Experience

Agent Runtime Platform

Agent Runtime Platform

Agent Runtime Platform

Enterprise Data & AI Management

Enterprise Data and AI Management

Why Open Insight?

Most Data Agent systems are designed to answer questions.

Open Insight is designed to achieve business goals.

Enterprises don't measure conversations.

They measure outcomes.

Built for Enterprise, Not Personal AI

Most AI Agents are designed for individual users. Each user installs, configures and manages their own AI environment.

Open Insight takes a different approach. The enterprise learns once. Everyone benefits.

Open Insight continuously learns from the enterprise โ€” its data, systems, business knowledge, workflows, skills, and expertise. This knowledge is centrally managed, governed, and made available to employees across the organization.

It runs on enterprise infrastructure, not personal laptops. No installation, no setup, no maintenance burden on employees. It works 24/7 โ€” monitoring systems, processing tasks, and collaborating with other AI agents across departments.

Roadmap

Component Shipped Next
ARP โ€” where AI employees run Multi-provider chat ยท agents & MCP ยท shared skills ยท self-hosted deploy Deeper ONE-PI integration ยท community agent & MCP templates ยท local browser extension
ONE-PI โ€” how AI employees think Virtual experts ยท shared skill repository ยท OpenAI-compatible agent API A2A collaboration ยท more experts
DMP โ€” what enterprises get Four learning engines ยท governed enterprise core (commercial) Deeper enterprise knowledge base ยท more enterprise connectors ยท deeper governance

A2A, redefined. Multi-agent systems let one framework orchestrate many agents. A2A in Open Insight is different: every person commands their own AI, and the collaboration that used to happen person-to-person now happens agent-to-agent โ€” with humans approving what matters.


Community & Enterprise Edition

OpenInsight follows an Open Core licensing model and is available in two editions:

Community Edition Enterprise Edition
Core Platform โœ“ โœ“
Deployment Self-hosted Self-hosted
Enterprise Features โ€” โœ“
License Not required Required

Community Edition is free to use and can be deployed without a commercial license.

Enterprise Edition includes additional enterprise and commercial capabilities and requires a valid License File issued by OpenInsight.

For more information, see LICENSE-COMMERCIAL.md.

System Requirements

Item Minimum Recommended
CPU 4 cores 16 cores
RAM 16 GB 32 GB
Disk 100 GB 500 GB

Prerequisites: Docker 24.07+, Docker Compose v2.26.1+

OS: Ubuntu 20.04/22.04, CentOS/RHEL 7.9, CentOS 8.6, RHEL 8.5

Quick Start

1. Clone the repository

git clone https://github.com/OpenInsightHQ/openinsight.git
cd openinsight

2. Initialize environment

bash init-env.sh

This generates env.sh from env.sh.example and automatically fills passwords and keys with random values.

3. Set the server IP

Edit the generated env.sh and change HOST_IP to the actual server IP:

vi env.sh
HOST_IP=192.168.1.100    # Your server IP

4. Pull images (online)

bash prepare.sh

Offline: run bash save-images.sh on a networked machine, copy docker-images.tar.gz to the target server, then run bash load-images.sh on the target server.

5. One-click install

bash install_all.sh

6. Post-install steps

Community Edition

No commercial license is required. Once installation is complete, you can access the platform directly.

  • DMP: http://<HOST_IP>:30080/dmp/
  • ARP: http://<HOST_IP>:33080/arp/

Enterprise Edition

Enterprise features require a valid License File. After installation, open the DMP management interface and activate the Enterprise License.

๐Ÿ“– For full details, see the Deployment Guide.

Configuration

All configuration is stored in env.sh, generated from env.sh.example by init-env.sh.

  • Secrets/Passwords: automatically generated randomly on first run
  • HOST_IP: the only setting you must change
  • External middleware: set USE_EXTERNAL_MYSQL/REDIS/MINIO=true to use external services
  • PI Agent LLM: configure OPENCODE_API_KEY and PI_MODEL after deployment

Directory Structure

openinsight/
โ”œโ”€โ”€ init-env.sh              # Environment initialization
โ”œโ”€โ”€ install_all.sh           # One-click installation
โ”œโ”€โ”€ uninstall_all.sh         # One-click uninstall
โ”œโ”€โ”€ prepare.sh               # Pull Docker images
โ”œโ”€โ”€ save-images.sh           # Export images for offline deployment
โ”œโ”€โ”€ load-images.sh           # Import images for offline deployment
โ”œโ”€โ”€ env.sh.example           # Configuration template
โ”œโ”€โ”€ common.sh                # Shared function library
โ”œโ”€โ”€ docs/                    # Documentation
โ”œโ”€โ”€ mysql/                   # MySQL + initialization SQL
โ”œโ”€โ”€ mongodb/                 # MongoDB
โ”œโ”€โ”€ redis/                   # Redis
โ”œโ”€โ”€ minio/                   # MinIO object storage
โ”œโ”€โ”€ meilisearch/             # Meilisearch full-text search
โ”œโ”€โ”€ searxng/                 # SearXNG + MCP
โ”œโ”€โ”€ codeinterpreter-api/     # Code Interpreter API
โ”œโ”€โ”€ pi-agent/                # ONE-PI agent service
โ”œโ”€โ”€ dmp/                     # DMP data management platform
โ”œโ”€โ”€ arp/                     # ARP agent platform
โ””โ”€โ”€ mongo-express/           # Mongo Express administration UI

Service Management

# Start / stop / restart a single service
cd <module> && docker-compose up -d
cd <module> && docker-compose down
cd <module> && docker-compose restart

# View all service statuses
docker ps --filter "name=openinsight-"

# One-click uninstall (keeps data)
bash uninstall_all.sh

Default Ports

Service Port
DMP 30080
ARP 33080
Code Interpreter API 8000

License

OpenInsight uses an Open Core licensing model.

  • Open-source components and materials covered by applicable open-source licenses are provided under their respective licenses, including the Apache License 2.0.
  • Enterprise features and other proprietary commercial capabilities require a valid commercial License File.

See LICENSE for the Apache License 2.0 and LICENSE-COMMERCIAL.md for commercial licensing information.

About

Enterprise Agent Operating System โ€” AI employees that learn your business and deliver outcomes. Self-hosted, open core.

Topics

Resources

Code of conduct

Security policy

Stars

47 stars

Watchers

1 watching

Forks

Releases

Packages

Contributors

Languages