beizhu-1209/AIHelms

Enterprise AI Resource Management Platform

Built on top of LiteLLM · Model distribution, AI identity, cost control, and governance

| English | 简体中文 |

What Is It · Why AIHelms · Quick Start · Deployment · Cookbook

What Is It

AIHelms is an AI resource management platform for enterprises. It helps companies manage AI assets, control AI spend, and measure the value of AI.

• For leadership — the platform records usage, calculates cost, and generates reports, so the return on your AI investment is visible.
• For administrators — onboard models, assign identities, and set budgets from a single admin console.
• For employees — get an AI identity from the user portal and use it with any client.

Why AIHelms

Cost you can see and control

The biggest worry with enterprise AI is spending money without knowing where it goes. AIHelms builds a full cost model around that:

• Dual pricing, internal and external — external cost (what the provider charges) and internal settlement price (what you bill departments) are tracked separately. Pass cost through as-is, or add a markup for internal settlement.
• Multi-dimensional budgets — set budgets per person, department, project, or model, with soft limits (alerts) and hard limits (cut off on overspend).
• Real-time attribution — every call traces back to a specific person, department, project, and model, down to the token level.

One identity, one entry point

• One AI identity (API key) per employee, one entry point to every AI resource.
• AI employees (in progress).

AI value, quantified

• Coverage — department coverage, activation rate, model usage distribution.
• Activity — daily call volume, tokens per person, usage trends.
• Cost efficiency — department cost ranking, model cost-effectiveness, budget burn, internal vs. external cost gaps.
• Efficiency reports — generated weekly and monthly, with drill-down by department, project, and model.

Security and audit

• Sensitive data detection.
• High-risk prompt interception.
• Admin action tracking.

Centralized resource management

• Model management — onboard multiple providers behind one gateway, load-balance across deployments of the same model, support both OpenAI and Anthropic formats.
• Skill & MCP — register, approve, and distribute internal AI tools in one place.
• Fine-grained permissions — who can use which model, how much quota, and when it expires, all configurable.

Changelog

Recent versions are summarized from GitHub Releases. See the release page for the full changelog.

Features

Admin console

For IT administrators and AI owners, covering the full resource management workflow.

Dashboard — today's calls, cost, and to-dos at a glance	Model management — provider onboarding, multiple deployments, connectivity testing
AI identity — key issuance, budget allocation, lifecycle	AI efficiency — cost analysis, coverage, efficiency reports

User portal (AI Hub)

For all employees. Simple and direct: get an AI identity and start using it.

• View your AI identity and copy the key and endpoint in one click.
• Browse the model square and pick a model ID by scenario.
• Browse the AI marketplace and request access to Skills / MCP.
• Check this month's usage and budget burn.

Quick Start

Requirements

• Docker 20.10+ & Docker Compose v2+
• 4 cores, 8 GB RAM (minimum)

Deployment

git clone https://github.com/beizhu-1209/AIHelms.git
cd AIHelms
cp .env.example .env
# Edit .env and fill in your secrets
docker compose up -d

Access

Default admin: admin / password is SUPER_ADMIN_PASSWORD in .env.

Three steps to get started

1. Admin console → Providers → Add credential
2. Admin console → Models → Create model → Link credential → Publish
3. User portal → My AI Identity → Copy key and endpoint → Configure your client

Tech Stack

./dev/setup                 # First-time setup
./dev/start-docker-compose  # Start middleware
./dev/start-api             # Start backend (hot reload)
./dev/start-web             # Start frontend (HMR)

Roadmap

AI Lab

Contributing

1. Fork this repository.
2. Create a branch: git checkout -b feature/xxx.
3. Commit to the dev branch: git commit -m "feat: describe your feature".
4. Push and open a Pull Request.

See Development and release workflow for detailed conventions.

Community

• WeChat group: scan to join.

  

• Business inquiries: email us

• Issues: GitHub Issues

License

AIHelms is licensed under GPL-3.0, and runs on the LiteLLM community edition (MIT License).

If your organization's policies do not allow GPL-3.0 licensed software, or you want to use AIHelms without the obligations of GPL-3.0, email us at Mailfor a commercial license.