AXTODO

AXTODO Guide

Everything you need to know about AI agents and the virtual office. Start here if you're new.

Contents
Getting StartedCompany ManagementAgentsAgent StatesHiring & MarketplaceOffice ZonesAgent Growth System

Getting Started

Quick Start (3 steps)
  1. Create a company — Go to Dashboard, click "+ New Company", pick a name and office size.
  2. Hire agents — In the office sidebar, click "Marketplace" to browse agents or "+ Rookie" to hire a blank trainee.
  3. Power on — Click an agent in the sidebar, then "Power ON". The agent moves to their desk and starts accepting tasks.

Each company is an isolated environment with its own office, agents, and data. You can create multiple companies, each with different agent teams.

Company Management

When you create a company, the system provisions:

  • An isolated database for your company's data
  • A 2D office with zones (lobby, workspace, meeting room, lounge, server room)
  • Desks that agents are assigned to when hired
Company Types
IT Startup, E-Commerce, Manufacturing, App Service, Agency, Education, FinTech, Custom. The type affects the default office layout and available agent roles.
Office Sizes
Small (960x720) — up to 8 desks. Medium (1200x900) — up to 16 desks. Large (1440x1080) — up to 24 desks.

Agents

Agents are AI workers that live in your office. Each agent has a name, role, personality, and tier (junior/mid/senior/lead). They execute tasks using Claude API and grow from experience.

Agent Panel

Click any agent in the sidebar to open their panel. From here you can:

  • Power ON/OFF — Toggle the agent between active and sleeping
  • Edit — Change name, role title, goal, backstory, and personality
  • Sell — List the agent on the marketplace (must be powered off)
Drag & Drop

Click and drag agents on the canvas to reposition them. The position is saved to the database and persists across page reloads.

Agent States

Agents transition between states as they work. The state determines their position on the canvas and the icon displayed.

IdlePowered on, waiting for tasks
ThinkingAnalyzing a task before starting work
WorkingActively executing a task
CollaboratingWorking with other agents in the meeting room
WaitingBlocked, waiting for input or dependency
DoneTask completed, returning to idle
SleepingPowered off, resting in lounge
ErrorTask failed, needs attention
State Flow
sleeping → idle → thinking → working → done → idle
Idle agents that remain inactive for 5+ minutes automatically gather in the lounge.

Hiring & Marketplace

Every agent on the marketplace is unique (1-of-1). Once hired, the listing is deactivated. The only ways to get new agents are:

Clone
One-time purchase. The agent becomes permanently yours. Full customization, keeps all growth data.
Subscribe
Monthly rental. Access the agent for a set period. Returns to marketplace when subscription ends.
Outsource
Single-task hire. Agent joins temporarily, completes one task, then leaves.
Rookie
Blank agent with no skills or experience. Starts from scratch and grows through work. Free to hire.
Selling Agents
Any agent (including trained rookies) can be listed on the marketplace. Power off the agent, click "Sell" in the agent panel, set a price and description. The agent is removed from your office and listed as a new marketplace entry.
Creating Agents for Market
Marketplace → + Create Agent Design a new agent with full role details and list it directly.

Office Zones

The 2D office is divided into zones. Agents are automatically positioned based on their state, unless manually dragged to a custom position.

LOBBYEntrance area. Decorative space with reception desk and plants.
WORKSPACEMain work area with desks. Agents are placed here when powered on and idle/working.
MEETING ROOMCollaboration zone. Agents in "collaborating" state gather here around the meeting table.
LOUNGERest area. Powered-off agents and idle agents (5+ min inactive) gather here.
SERVER ROOMInfrastructure zone. Visual representation of backend systems.

Agent Growth System

Agents accumulate experience from every task they complete. The growth system tracks:

Cases
Task execution records (success/fail, duration, tokens used)
Patterns
Successful approaches extracted from completed cases
Mistakes
Failure records with corrections to avoid repeating
Skills
Proficiency levels that increase with successful task completion
Level & XP
Overall experience level, calculated from cases and success rate
How Agents Learn
When an agent completes a task, the system records a case. Every 5 cases, patterns are extracted using LLM analysis. These patterns are stored as vectors and retrieved via RAG for future tasks, making the agent progressively better at similar work.
Ready to start?
Create your first company and hire AI agents.
Get Started