# Voicetree > An infinite canvas for orchestrating coding agents. Flow-state context engineering for developers. Voicetree is a spatial IDE for multi-agent orchestration. It organizes project context as an interactive graph of markdown files and terminals, letting developers visually manage multiple AI coding agents working in parallel. ## Core Features - Visual agent orchestration: every agent sits next to its task, planning docs, and progress updates - Task decomposition: break complex work into isolated subtasks with execution order and parallelization - Transparent subagent spawning: agents spawn visibly on the canvas in their own terminals - Context preservation: no context loss between sessions, spatial navigation of project state - Local-first: everything stored on-device as markdown files - Token efficiency: context pruning leads to ~60% fewer input tokens ## Links - Website: https://voicetree.io - Docs: https://voicetree.io/docs/How-it-works - Getting Started: https://voicetree.io/docs/Getting-started-with-obsidian-vaults - GitHub: https://github.com/voicetreelab/voicetree - Discord: https://discord.gg/voicetree - Contact: hello@voicetree.io - Full content: https://voicetree.io/llms-full.txt ## Pages ### Core - About Us: https://voicetree.io/about Meet the creator of Voicetree. Built by Manu, a software engineer from Sydney, to solve the context management problem across tasks and AI coding agents. - Voicetree for Context Engineering: https://voicetree.io/context-engineering An infinite canvas for orchestrating coding agents. You architect, agents build. Spatial navigation, transparent orchestration, and divide-and-conquer agent management. - Download Voicetree: https://voicetree.io/download Download Voicetree for macOS (Apple Silicon & Intel), Windows, and Linux. Install via Homebrew or direct download. - Voicetree - A Graph View You Actually Work Within: https://voicetree.io The graph holds your memory so you don't have to. Agents spawn agents. Watch from above, zoom into where you are needed. ### Documentation - Getting Started with Obsidian Vaults: https://voicetree.io/docs/Getting-started-with-obsidian-vaults Quick start guide for using Voicetree with Obsidian vaults. Install, open your vault, and start creating nodes. - How Voicetree Works: https://voicetree.io/docs/How-it-works Technical overview of Voicetree's voice-to-graph pipeline and local-first architecture. Learn how speech becomes structured knowledge graphs using vector search and LLMs. - Privacy Policy: https://voicetree.io/docs/Privacy-Policy Voicetree privacy policy explaining how we handle your data, what we collect, and our commitments to data protection. We don't sell your data or use it for advertising. - Voicetree for Tree Style Terminal Tabs: https://voicetree.io/docs/Voicetree-for-Tree-Style-Terminal-Tabs Use Voicetree as a graphical interface for organising many terminal tabs with one-click native terminals and attached notes. ### Blog - Agentic Engineering Guide: Talk, Plan, Decompose, Review: https://voicetree.io/blog/agentic-engineering-guide 18 months and 990k LOC of agentic development distilled into a practical engineering loop. Inspired by Japanese train drivers' Point & Call method and functional programming's type-first design. - The Complexity Threshold: Why Your AI Agent Produces Gold or Garbage: https://voicetree.io/blog/complexity-threshold LLMs have a sharp complexity threshold - below it they produce magic, above it pure slop. Learn to recognize the "Claude ego spiral" anti-pattern and use practical complexity budgeting to keep your AI agents productive. - Why Functional Architecture Makes AI Agents Actually Work (Lessons from a 20-Hour Flight): https://voicetree.io/blog/functional-architecture-for-agentic-frontend How designing frontends with functional architecture transforms agent-generated code from spaghetti to composable, testable systems. - History of Voicetree: https://voicetree.io/blog/History The journey of building Voicetree from a productivity tool for knowledge workers to an AI agent orchestration platform. Learn about our breakthroughs in context engineering and agent coordination. ### FAQ & Questions - Frequently Asked Questions: https://voicetree.io/blog/more/FAQ Common questions about Voicetree, the interactive graph-view for orchestrating coding agents. Learn about context engineering, voice mode, agent orchestration, supported coding agents, API cost reduction, and how Voicetree improves AI coding workflows. - How do you manage multiple AI coding agents running in parallel?: https://voicetree.io/blog/more/questions/managing-multiple-ai-agents Learn how to track and coordinate multiple AI coding agents working simultaneously without losing context or creating conflicts. - How do you run AI coding agents overnight without them going off track?: https://voicetree.io/blog/more/questions/overnight-autonomous-workflows Strategies for unattended autonomous agent runs, covering context preservation, Ralph Wiggum loops, file-based state, and how Voicetree makes morning review fast. - Why was Voicetree built?: https://voicetree.io/blog/more/questions/voicetree-origin-story Voicetree was built to let you explain work out loud while AI agents handle implementation. Focus on architecting while coding agents do the painful parts. - How does Voicetree compare to alternatives?: https://voicetree.io/blog/more/questions/voicetree-vs-alternatives How Voicetree compares to Gas Town, GSD, Ralph Wiggum, GraphRAG, Superpowers, and plain terminal workflows for managing AI coding agents.