Key features
AI Priority Manager: Automatically scans open tasks, identifies blocking items (e.g., 'Task 27 · Development Project Blocked'), proposes status change suggestions (e.g., 'Move Task 27 to Blocked'), and submits them to the AI Inbox for manual approval;
Intelligent Task Decomposition Skill (Intake/brief.parse): Based on natural language task briefs (e.g., 'Plan the spring open day: venue, invitations and signage'), it automatically generates a list of subtasks with estimated durations (e.g., 'Book the venue', 'Send invitations', 'Order signage') and flags any outstanding questions, supporting dry-run previews.
Use cases and audience
Organizations that require autonomous control over data and processes—such as software engineering teams, clinical operations teams, procurement and supply chain teams—are well-suited to use this solution. Applicable scenarios include project management, ERP extensions, defect triage, customer support ticket escalations, and more.
- Automated project prioritization: AI agents automatically analyze incomplete tasks and suggest adjustments, such as moving tasks to the blocked list.
- Intelligent task decomposition: Generates subtasks based on task briefs, estimates durations, and raises questions needing clarification.
- Supplier Order Exception Monitoring: Identify purchase orders with delayed deliveries through natural language queries, and pinpoint the specific supplier along with the number of overdue days.
Natural language queries (e.g., 'Which supplier orders are late this week?'), task briefing texts, project board statuses, calendar data, and synchronized data from external systems (GitHub/GitLab/Slack/Teams/Zapier/Google Calendar).
Operation suggestions proposed by AI agents (e.g., moving tasks, creating subtasks, adding comments), structured summaries (e.g., project pulse reports, CEO reports), lists of changes awaiting approval (displayed in the AI Inbox), and dry-run preview results.
Frequently asked questions
What data can the AI agent access?
The AI agent can only read the projects and tasks that the current user is authorized to open, without accessing other projects (as stated in the original text: "It sees only the projects you can open." and "Answers come only from the projects you can open.")
Does it support connecting to custom large models?
Support: Administrators can configure connections to their own AI models on the server side, including Claude, ChatGPT, Claude Code, and others, integrated via the MCP protocol (as in the original text: 'Connect your own AI' and 'ClaudeChatGPTClaude CodeAlianHub MCPacts as you')
How to use
- 01
Visit live demo.alianhub.com to experience it directly—no installation required.
- 02
Alternatively, run it on a local server: download docker-compose.yml, set the JWT_SECRET environment variable, and execute `docker compose up -d`.
- 03
Open http://localhost:4000 to enter the workspace.
- 04
The AI agent only reads project data that the user has permission to access, and all operations require manual approval.
Pricing and availability
Permanently free self-hosting; licensed under the AGPL-3.0 open-source license; no per-seat pricing model; AI agent runtime usage is counted against a monthly quota (as shown on the demo page: '$0.00 of $36 this month'), but the official website does not specify the exact billing metrics—whether by word count, character count, or API calls—as well as commercial licensing terms.
- Pricing source
- https://alianhub.com/
- Pricing last updated
- 2026-10-27
Prices and limits may change. Check the official website for current terms.
Official website & sources
Updated on 2026-10-27
