Rabbitholes

About Rabbitholes

Most AI chat tools turn into a mess after 10 messages. Context gets muddled. You're scrolling up to remember what you asked three prompts ago. You lose track of which question led to which answer.

Rabbitholes takes a different approach. It's an infinite canvas AI tool where each conversation is a node. You can branch conversations, connect multiple AI models to different chats, and cherry-pick context from one node to feed into another. No more reprompting the same background information 15 times. The canvas layout means you can see all your AI conversations spatially, not buried in a linear chat history that forces you to scroll endlessly.

It's a desktop app that costs $89 for one year of updates or $249 for lifetime updates. You bring your own API keys from providers like OpenAI, Anthropic, Google, and Perplexity. The conversations stay stored locally on your device, not on someone else's server.

Try Rabbitholes and transform how you work with AI.

---

What is Rabbitholes?

Think of it as a spatial workspace for AI conversations instead of a traditional chat interface.

Each chat lives in its own node on an infinite canvas. You can create multiple nodes, position them however you want, and connect them with lines to show relationships. When you need context from a previous conversation, you literally draw a connection between nodes instead of copying and pasting text. The tool pulls that context automatically.

You can switch between AI models mid-conversation or run multiple models on the same prompt simultaneously. Want to see how Claude and GPT-4 respond to the same question? Put them side by side. The node-based structure makes branching conversations simple - if an AI response takes your thinking in two different directions, split the node and explore both paths without losing either thread.

It works with major providers (OpenAI, Anthropic, Google, Perplexity, and Groq) and supports Ollama for running local models. You can attach files, websites, images, PDFs, Excel sheets, and YouTube videos to nodes as source material. The system prompt, temperature, and top_p settings are all adjustable if your chosen model supports it.

---

Who is Rabbitholes For?

This makes sense for people who have complex, multi-threaded conversations with AI models.

Researchers who need to explore multiple angles of a question simultaneously will appreciate the branching. Instead of managing 12 browser tabs with different ChatGPT conversations, you can see all your research threads on one canvas. Connect related ideas visually. Pull context from your literature review node into your analysis node without retyping summaries.

Students working on thesis projects or complex assignments benefit from the spatial layout. You can keep your research in one area of the canvas, your outline in another, and your drafting conversations in a third section. When your professor's feedback requires rethinking a section, branch from that point instead of starting over.

Knowledge workers who context-switch constantly - project managers, strategists, consultants - can organize conversations by client or project. The 2-device limit means you can run it on your work laptop and home desktop, reassigning devices when you upgrade.

It's overkill if you only ask AI simple, one-off questions. The spatial canvas and node connections matter when you're building on previous conversations, not when you're just asking for recipe ideas.

---

Rabbitholes Pros and Cons

The value proposition depends heavily on how much you use AI tools. If you're already paying for ChatGPT Plus and Claude Pro ($44/month combined), the $89 annual plan pays for itself in 2 months. But add in API costs, and you need to actually use the spatial features enough to justify the learning curve and setup hassle.

---

Rabbitholes Features: Canvas, Nodes, Multi-Model & File Support

Node-Based Chat System

Every conversation is a separate node on the canvas. You can position them anywhere, zoom in and out, and organize them spatially however makes sense to you. The branching capability means when a conversation reaches a decision point, you can split it into two nodes and explore both options. Researchers will use this constantly - testing different hypotheses without losing work. The downside is that managing dozens of nodes can get visually cluttered if you don't stay organized.

Multi-Model Support and Switching

You can switch between AI models mid-conversation or create multiple nodes with different models answering the same prompt. The tool supports OpenAI, Anthropic, Google, Perplexity, Groq, and custom providers. Want GPT-4's reasoning on a problem and Claude's writing style on the response? Set up two connected nodes. The Ollama integration lets you run local models if you have the hardware. This is useful for completely offline work or when you're testing open-source models.

Context Connection Between Nodes

Instead of copying and pasting background information into every new chat, you connect nodes. Drag a line from your research node to your writing node, and Rabbitholes feeds that context automatically. This solves the "context pollution" problem where you keep repeating yourself to the AI. The connection is visual, so you can see which conversations are building on which source material.

File and Source Attachment

You can attach these formats to nodes:

• PDFs, Word docs, PowerPoint presentations
• Excel spreadsheets
• Images (PNG, JPG)
• Websites
• YouTube videos

The AI can reference these sources when generating responses. If you're analyzing a research paper, attach the PDF to the node. If you're summarizing a meeting, attach the recording. The tool doesn't list limits on file sizes, which could be a problem if you try uploading a 500MB video.

Advanced Model Controls

You can adjust system prompts, temperature, and top_p settings for each conversation. This matters when you need consistent outputs (lower temperature) versus creative brainstorming (higher temperature). The controls are subject to what each model provider supports - some models don't expose all these settings. Technical users will appreciate this level of control. Casual users might ignore it entirely.

Export and Local Storage

Conversations save locally on your device, not in the cloud. You can export your entire canvas as JSON or markdown files. Good for backups, sharing specific conversations with colleagues, or migrating to other tools. The local storage means if your hard drive fails without backups, you lose everything. There's no automatic cloud sync option.

Start using Rabbitholes to organize your AI workflow spatially.

---

Rabbitholes vs Alternatives: Pricing & Feature Comparison

ChatGPT Plus and Claude Pro are simpler. You pay monthly, log in on any device, and start chatting immediately. No API keys, no setup. But you're locked into one model family and the linear chat interface gets messy with complex projects.

Rabbitholes wins on flexibility and long-term cost if you're already juggling multiple AI subscriptions. The $89 annual plan costs less than 5 months of ChatGPT Plus. Add in the ability to compare models side-by-side, and it's worth it for people doing serious AI-assisted work. But the API costs add up - running GPT-4 extensively through the API could cost $30-50/month on top of the Rabbitholes license.

The spatial canvas is the real differentiator. If you find yourself constantly scrolling through chat histories trying to remember what you discussed 20 messages ago, the node system solves that. If you're fine with linear chats, you're paying for a feature you won't use.

Skip Rabbitholes if you primarily use AI on mobile devices. The desktop-only limitation means you can't pull out your phone to quickly ask a question. ChatGPT and Claude's mobile apps win there.

---

Rabbitholes Pricing: Plans & Cost Breakdown

Both plans require you to bring your own API keys, which means ongoing costs beyond the license fee.

The $89 plan makes sense if you want to test whether the spatial workflow actually improves your AI usage. After a year, you'll know if it's worth upgrading to lifetime or if you'd rather stick with standard chat interfaces. The $249 lifetime deal is cheaper than 13 months of ChatGPT Plus alone - if you're already paying for multiple AI subscriptions, it pays for itself in under a year.

But here's what people miss: the API costs. OpenAI charges $0.01 per 1,000 tokens for GPT-4o input (as of 2025). If you're having long, complex conversations daily, that's easily 50,000-100,000 tokens, or $0.50-1.00 per day. Multiply by 30 days and you're at $15-30/month in API fees on top of the Rabbitholes license. Claude and other providers have similar pricing structures.

Compared to tools like Notion AI ($10/month per user) or Reflect ($10-30/month), Rabbitholes is cheaper long-term but requires more technical setup. You're trading convenience for cost savings and flexibility.

The 2-device limitation is annoying if you regularly use more than two computers. You can reassign devices, but it's an extra step. Unlimited device support would justify a higher price tier, but that doesn't exist yet.

---

Is Rabbitholes Worth It? Honest Review

I've been using Rabbitholes for several weeks now, and honestly, this might be the superior way to work with AI models. It's far more effective than a traditional chat interface, especially when I'm working on projects that branch into multiple directions. The amount of control you have is incredible - being able to switch models mid-conversation or run the same prompt through Claude and GPT-4 simultaneously saves so much time.

What I didn't expect was how much the spatial canvas would change my workflow. I'm someone who typically has 15 browser tabs open with different AI conversations, constantly losing track of which tab contained that important response from yesterday. With Rabbitholes, I can see all my related conversations on one screen, positioned in a way that makes sense to me. Research in one corner, drafting in another, editing feedback in a third area.

The app design is so beautiful that I actually enjoy using it. That sounds trivial, but when you're spending hours working with AI, interface aesthetics matter. I'm also surprised it's a local app rather than browser-based. I much prefer that. It feels faster, more stable, and I'm not dealing with browser memory issues when I have 20 nodes open on my canvas.

The main drawback I've found is the API cost tracking. I'm still figuring out exactly how much I'm spending on API calls each month since that's separate from the Rabbitholes license. It would be helpful if the app showed real-time cost tracking as I use different models.

---

Rabbitholes Review: Final Verdict

Rabbitholes is worth the $89 investment if you're already spending $20-40 monthly on AI subscriptions and finding linear chat interfaces limiting. The spatial canvas and node system genuinely solve real problems - context loss, scattered conversations, and the inability to explore multiple paths simultaneously. Power users doing research, writing, or strategic work will see immediate value.

It's not for casual users who ask AI occasional questions or people who primarily use AI on mobile devices. The desktop-only limitation and API key setup create friction that simple web-based tools don't have. If you're happy with ChatGPT's standard interface and rarely need to reference previous conversations, save your money. But if you're juggling multiple AI conversations daily and losing track of context, the node-based approach will feel like a significant upgrade. The lifetime plan at $249 is the better deal for anyone confident they'll use spatial AI workflows long-term.

Get started with Rabbitholes and experience AI on an infinite canvas.

---

FAQ

Does Rabbitholes include AI access, or do I need my own API keys?

You need to bring your own API keys from providers like OpenAI, Anthropic, Google, and Perplexity. The tool doesn't include AI access - it's a workspace for organizing and managing conversations with models you're already paying for separately.

Can I use Rabbitholes on mobile devices?

No, it's desktop-only. The app works on Mac, Windows, and Linux, but there's no mobile version. You're limited to the 2 devices associated with your license.

What happens to my conversations if I don't renew after one year on the $89 plan?

You keep full access to the app and your existing conversations. You just don't get new feature updates or bug fixes released after your year expires. Everything you've already created remains accessible.

Does the app work offline, or do I need internet for the AI models?

If you're using cloud-based models (OpenAI, Anthropic, etc.), you need internet to send requests to their APIs. But if you're running local models through Ollama, you can work completely offline. Your conversations and canvas are stored locally either way.

How much do API costs typically run if I'm using the tool daily?

It varies wildly based on which models you use and how much you chat. GPT-4o costs about $0.01 per 1,000 input tokens. Heavy daily users might spend $20-50/month on API calls. Claude and other providers have similar pricing structures. Track your usage through each provider's dashboard to avoid surprises.