Built for teams managing multiple AI agents

Make AI agents easier
to deploy and manage

Built on OpenClaw and designed for containerized deployment. Manage multiple agents, knowledge bases, messaging integrations, monitoring, and usage costs from one clean control panel.

Manage multiple agents in one place Isolated runtime for each agent Feishu / WeCom / DingTalk integrations Built for long-term operations
Book a demo Explore features
15+
core modules
100%
container isolation
30s
health-check interval
24/7
automated monitoring

Everything you need in one management layer

From agent creation to monitoring and operations, the platform covers the full lifecycle of enterprise AI agents.

Live status dashboard

See all agent states at a glance, with real-time CPU and memory trends to keep operations clear and visible.

Multi-agent management

Each agent runs in its own container, with one-click create, start, stop, restart, and bulk actions across agents.

Flexible model routing

Connect OpenAI, Anthropic, DeepSeek, and compatible self-hosted APIs, all behind one routing layer.

Skill marketplace

Install new capabilities from a skill library or upload custom skill packages to extend each agent.

Conversation history

Keep complete chat records for every agent and review user interactions in a structured timeline.

Agent customization

Define persona, role, naming, and system prompts so each assistant fits its business purpose.

Messaging integrations

Support Feishu, WeCom, DingTalk, Slack, official WeChat accounts, and other team communication channels.

Usage and cost tracking

Track token usage and estimated costs by agent and by user, with clear trends over time.

Security and compliance

JWT auth, encrypted API key storage, and audit logging help teams operate safely and with traceability.

Knowledge base management

Each agent can keep its own knowledge base, with drag-and-drop uploads and previews for text, images, and PDFs.

One-click cloning

Clone agent settings, knowledge files, and skill bindings to launch a matching agent in seconds.

Cost estimation

Set model pricing once and let the platform automatically estimate spend from actual usage.

Disk monitoring

Monitor storage usage across agents, backups, and skills, with warnings before space becomes a problem.

Data export

Export chat logs and audit records as CSV for reporting, archiving, or further analysis.

The questions customers ask most

These are the questions teams usually ask before moving forward, so you can quickly judge whether the platform fits your use case.

What does a multi-agent platform actually mean?

It is not just one chatbot. It is a control layer for running many AI agents across multiple teams, roles, and workflows from one platform.

How is HaiHe Agent different from a normal AI assistant?

Most AI assistants serve one narrow use case. HaiHe Agent is designed for enterprise operations, with centralized control, observability, integrations, and long-term maintainability.

Can it be deployed on our own infrastructure?

Yes. The platform can run on your own cloud servers or internal infrastructure, and it can connect to your existing model gateway if needed.

Does it support Feishu, WeCom, and DingTalk?

Yes. It fits Feishu bots, WeCom assistants, DingTalk workflows, and can be extended to Slack, official WeChat accounts, and API-based integrations.

How do we handle permissions and cost control?

You can track usage by agent, user, and team, while assigning separate knowledge bases, model routes, and access rules for different business units.

What business scenarios is it best for?

It works well for support teams, sales enablement, internal knowledge assistants, content teams, legal and finance workflows, and engineering operations.

One control layer for all your agents

Deployments, knowledge bases, messaging channels, live status, and cost tracking all live in one place.

Container isolation with bulk controls

Manage all business agents from one server and one routing layer, while keeping each agent isolated in its own runtime.

  • Create new agents in one click, with ports, folders, and configs assigned automatically
  • Start, stop, or restart multiple agents in bulk
  • Clone an existing agent with its config, knowledge base, and skills
  • Connect Feishu, WeCom, DingTalk, Slack, and other messaging channels
  • Share one model route across all agents or bind different models where needed
4 agents selected
Start Stop Restart
Customer Support GPT-4o Running
Document Analyst Claude Running
Code Assistant DeepSeek Degraded
Translation Engine Qwen Stopped

Give each agent its own domain knowledge

Dedicated knowledge bases make each agent more accurate and much more useful in its own field.

  • Independent knowledge bases for each agent
  • Drag-and-drop uploads with no workflow friction
  • Preview text, images, and PDFs directly in the interface
  • Copy the full knowledge set when cloning an agent
📄
Product Guide.pdf
2.4 MB
Preview
📝
FAQ.md
18 KB
Preview
📊
Pricing.xlsx
156 KB
Download
🖼️
Architecture.png
340 KB
Preview

Every conversation stays traceable

Session-based history keeps context intact and makes reviews far easier for operations teams.

  • See full conversations in chronological order
  • Filter system noise and keep only useful interactions
  • Parse Feishu message formats automatically
  • Preserve history even when active context is compacted
  • Export to CSV for reporting and archiving
Help me draft this week's progress update.
Absolutely. What were the main things you completed this week?
We finished the user management module and fixed 3 production bugs.
Got it. Here is a clean summary:

Weekly Update
1. Completed the user management module
2. Resolved 3 production issues...

Live monitoring with proactive alerts

Visual dashboards show overall health, while failures trigger alerts automatically so teams can react earlier.

  • CPU and memory trends across 24h, 7d, and 30d windows
  • Health checks every 30 seconds with automatic webhook alerts
  • Cooldown rules to avoid alert storms
  • Disk tracking split across agents, backups, and skills
  • Automatic backups with one-click recovery
12
Running
98%
Uptime
CPU Trend (24h)
Code Assistant Running → Degraded · Feishu alert sent

Container-based microservice architecture

Each agent runs in its own isolated container, keeping the system clean, resilient, and scalable.

Access Layer
Messaging Integrations
Web Admin Panel
API Layer
Management Layer
HaiHe Agent Control Layer
Runtime Layer
Agent 1
Docker Container
Agent 2
Docker Container
Agent 3
Docker Container
Agent N
Docker Container
Model Providers
OpenAI
Anthropic
DeepSeek
Self-Hosted Models

Start with the use cases that prove value fastest

These are the kinds of workflows where teams usually see clear value quickly.

E-commerce / online education support

A dedicated support agent for each store or course

A company running five online stores can deploy one support agent per store, each with its own product manuals and policies. Customers ask questions in Feishu, agents reply instantly, and management sees usage and cost by store.

Store A Support Store B Support Store C Support After-sales Specialist

Law firms / finance advisory

Separate agents by expertise with clean data boundaries

A mid-sized law firm can run separate agents for labor law, contract review, and IP. Each agent keeps its own regulations and case files, while all conversations remain auditable for compliance.

Labor Law Advisor Contract Review IP Assistant

Marketing / content teams

Different brand voices, different agent personas

An agency managing multiple brands can assign one content agent per brand, each trained on its tone and best-performing references. Teams can use different agents for social posts, long-form assets, or SEO work.

Social Copywriter Newsletter Editor Whitepaper Writer

Engineering internal tools

Review, docs, and ops assistants each with a clear role

A 20-person engineering team can deploy separate agents for code review, API docs, and infrastructure support. New hires ask the agents instead of interrupting senior teammates, and answers stay grounded in internal docs.

Code Review API Docs Ops Assistant

Real estate / local services

One AI receptionist per branch, managed centrally

A real estate chain can give each branch its own agent with branch-specific listings and scripts. Customers get fast first responses, while headquarters tracks performance and costs by location.

Chaoyang Branch Haidian Branch Fengtai Branch +5 more

Security and operations built in

JWT Authentication

JWT-based auth protects the admin surface and secures API access across the platform.

Secret Encryption

API keys are stored with encryption, reducing exposure risk even if the database is compromised.

Audit Logs

Critical actions are recorded automatically and can be filtered or exported for reviews.

Backup & Recovery

Combine scheduled and manual backups, with database dumps and config archives ready for recovery.

Feishu Alerts

When agent status changes unexpectedly, alerts can be pushed to Feishu with cooldown controls.

Container Isolation

Each agent runs in its own Docker container, so faults and resource pressure stay isolated.

Start lean, then scale as needed

Suitable for pilot projects, small teams, and formal production use. Model API and infrastructure costs are billed separately.

Starter
A good fit for a first team or a few focused use cases
$399 /month
  • Up to 3 agents
  • Core knowledge base features
  • Feishu / WeCom / DingTalk integrations
  • Basic monitoring
Contact us
Enterprise / Private Deployment
For multi-team environments and long-term stable operations
Custom /quoted
  • Unlimited agents
  • Private deployment on your own servers
  • Custom integrations and access controls
  • Dedicated operations support
Contact us
Note: listed prices are platform service fees in USD. Model usage charges and infrastructure costs are billed separately based on your setup.

If you are bringing AI into real team workflows, let's talk

See a live demo first, then decide what deployment and integration path fits your team best.

Book a demo View pricing
Stack: FastAPI + React + Ant Design + PostgreSQL + Docker

WeChat Contact

Add us on WeChat and mention “HaiHe Agent” to book a demo or discuss your setup.

jianghebtc
Reply within 24 hours on business days