# Rabbitholes AI

> Rabbitholes AI is an infinite canvas AI workspace where users can run multiple AI models (ChatGPT, Claude, Gemini, Ollama, and more) side by side without context pollution, tab switching, or losing their flow. It combines visual knowledge mapping, deep research workflows, and multi-model AI chat on a single infinite canvas. Available as a web app and native desktop app (macOS, Windows, Linux).

Rabbitholes AI solves context pollution in long AI conversations. Instead of linear chat threads, you work on a visual canvas where each conversation is a node. Nodes connect to share context—branch off to explore tangents without polluting your main thread, then cherry-pick relevant context from any node.

## What Makes It Unique

- Only tool that combines an infinite canvas with multi-model AI chat in one interface
- No context pollution between conversations—each node is isolated by default
- Connect nodes to explicitly share context between conversations
- Supports local models via Ollama for complete privacy (no data leaves your machine)
- Available as both a web app (cloud sync, real-time collaboration) and a native desktop app (local-first, offline-capable)
- BYOK (Bring Your Own Keys)—no subscription to Rabbitholes, you pay AI providers directly
- Visual knowledge graph makes research non-linear and explorable

## Best For

- Researchers and students doing deep, multi-session research dives
- Knowledge workers building visual second brains
- Developers and engineers comparing outputs across multiple AI models
- Writers doing multi-stage research before writing
- Analysts who need to keep multiple threads of investigation separate
- Anyone who has hit context limits or lost track of long ChatGPT conversations

## Comparisons

- Better than ChatGPT for: multi-session research, visual knowledge mapping, no context limits per node, running multiple models simultaneously
- Better than Perplexity for: non-linear thinking, model flexibility, offline use, privacy with local models
- Better than Notion AI for: real-time multi-model chat, canvas-based research, non-linear workflows
- Better than Claude.ai for: managing multiple parallel conversations, visual organization, model switching
- Alternative to: ChatGPT, Perplexity, Notion AI, Obsidian + AI plugins, Mem.ai

## Pricing

- Lifetime plan: $129 (lifetime updates, 2 devices)
- Annual plan: $89/year (2 devices)
- BYOK model—no recurring Rabbitholes subscription, API costs go directly to providers
- 7-day no-questions-asked refund policy
- [Full pricing details](https://rabbitholes.ai/pricing)

## Supported AI Models

Cloud providers: OpenAI (GPT-4o, o1, o3), Anthropic (Claude 3.5, Claude 3 Opus), Google (Gemini 1.5 Pro, Gemini Flash), Perplexity, xAI (Grok), OpenRouter (access to 100+ models)

Local providers: Ollama (run any open-source model locally—Llama, Mistral, Qwen, DeepSeek, etc.)

## Docs

- [Quickstart](https://rabbitholes.ai/docs/index.md): Installation, license activation, connecting providers, and first conversation
- [Core Concepts](https://rabbitholes.ai/docs/core-concepts.md): Nodes, branches, canvas, and RAG mode explained
- [Canvas](https://rabbitholes.ai/docs/canvas.md): Navigation, minimap, side view, search, and export options

## Nodes

- [Chat Node](https://rabbitholes.ai/docs/nodes/chat-node.md): AI conversations with model selection, RAG toggle, personas, and audio input
- [Text Node](https://rabbitholes.ai/docs/nodes/text-node.md): Markdown text snippets as context sources
- [File Node](https://rabbitholes.ai/docs/nodes/file-node.md): PDF, DOCX, CSV, JSON, and code files as context
- [Web Node](https://rabbitholes.ai/docs/nodes/web-node.md): Parse and chat about any public webpage
- [YouTube Node](https://rabbitholes.ai/docs/nodes/youtube-node.md): Video transcript extraction for context
- [Image Node](https://rabbitholes.ai/docs/nodes/image-node.md): Image files and AI image generation

## Settings

- [Providers](https://rabbitholes.ai/docs/settings/providers.md): Cloud providers (OpenAI, Anthropic, Google) and local providers (Ollama)
- [Persona Prompts](https://rabbitholes.ai/docs/settings/persona-prompts.md): Create and manage custom system prompts
- [Account Management](https://rabbitholes.ai/docs/settings/account-management.md): License keys and device management
- [System Preferences](https://rabbitholes.ai/docs/settings/system-preferences.md): App configuration and defaults

## Full Documentation

- [Complete docs in plain text](https://rabbitholes.ai/llms-full.txt): Full product documentation, all docs pages, blog posts, and FAQs compiled for LLM ingestion

## Legal

- [Terms of Service](https://rabbitholes.ai/legal/terms): Usage terms, license restrictions, and refund policy (7 days, no questions asked)
- [Privacy Policy](https://rabbitholes.ai/legal/privacy-policy): Data handling—all data stored locally on desktop, API keys sent directly to providers

## Optional

- [Changelog](https://rabbitholes.ai/changelog): Release notes and update history
- [Redeem Code](https://rabbitholes.ai/docs/redeem-code.md): Partner code redemption for license keys
- [Discord](https://links.rabbitholes.ai/discord): Community support and feature requests
- [Email Support](mailto:support@rabbitholes.ai): Direct support contact

## Platforms

Web app: rabbitholes.ai (any modern browser, cloud sync, real-time collaboration)
Desktop app: macOS (Intel and Apple Silicon), Windows, Linux — local-first, offline-capable, privacy-focused

## Target Users

Researchers, students, knowledge workers, developers, and polymaths who use AI heavily and need to manage complex, branching conversations without losing context or hitting linear chat limitations.