What is IO?

IO is a self-hosted AI orchestrator daemon that acts as your personal companion — always running, accessible via Telegram and a web dashboard. It manages specialized squads of AI agents that handle project work autonomously.

Core Concepts

Io (The Orchestrator)

Io is your main point of contact. It:

• Answers questions directly when no squad is relevant
• Delegates project-specific work to the appropriate squad
• Manages a personal wiki for long-term memory
• Runs scheduled tasks via a cron engine
• Routes messages to the optimal model tier (fast/standard/premium)

Squads

A squad is a team of AI agents assigned to a single repository. Each squad has:

• Team Lead — Plans objectives, assigns tasks, picks models
• QA Agent — Reviews work with veto power (max 3 revision cycles)
• Specialized Agents — LLM-generated roles based on your codebase (e.g., Frontend Dev, API Engineer, DevOps)

Execution Pipeline

When work is delegated to a squad:

Objective → Team Lead Plans → Tasks Assigned
  → Agents Execute in Parallel (git worktrees)
    → Review Meeting
      → QA Approval/Rejection
        → PR Created → Inbox Notification

Design Principles

• No universes or characters — Agents have roles, not personas
• One squad per repo — Clear ownership, no cross-squad coordination
• Human oversight — QA gates, inbox notifications, configurable PR modes
• Smart model routing — Use the cheapest capable model for each task
• Always-on — Daemon runs persistently, accessible from anywhere via Telegram or web