Welcome to the docs
What AI Context Orchestrator does, and where to start.
Updated September 25, 2026 · 1 min read
What it is
AI Context Orchestrator is a visual Canvas for designing multi-agent AI workflows. You draw the architecture: a Root Orchestrator that delegates to Sub-Agents, each with its Skills (tools), References (context) and Connectors (access to external tools).
The app turns that drawing into two things: a Blueprint you download for Claude Code (CLAUDE.md, one file per Skill, .mcp.json), and runs you can watch live in the execution console.
Start here
Co-Pilot: from a sentence to a Canvas
Describe a process; approve the design; the Canvas opens built.
Quickstart
From an empty Canvas to a validated, simulated and exported workflow.
Get the most out of the Free plan
What's included, and how to build real workflows without paying.
Prompt Engineering on the Canvas
Behavior rules and Skill descriptions that route well.
Integrations and the Vault
Connect tools through MCP with encrypted credentials.
Voice commands and dictation
Drive the Canvas and write prompts by voice.
Glossary
The shared vocabulary, from AST to Webhook.
How the pieces fit
- Design on the Canvas. The Pre-Flight panel checks every node while you type.
- Simulate a run for free: scripted events show how the workflow would execute, with no API call.
- Run Live with your own Claude API key (BYOK), kept encrypted in the Vault.
- Export the Blueprint and keep working in Claude Code.