Skip to main content
This guide walks you through your first NodePad session from beta sign-up to your first forked conversation. Each step builds on the last, so you’ll understand the canvas before you start experimenting on your own.
You need a NodePad beta invite before you can log in. If you haven’t requested access yet, visit node-pad.com/beta and submit your email. See Beta access for details on how the invite waves work.
1

Join the beta and log in

Go to node-pad.com/beta and enter your email to request access. When your invite arrives, follow the link in the email to create your account and log in. You’ll land directly on a blank canvas — no setup required.
2

Send your first message

Click anywhere on the empty canvas to create your first chat node. A text input appears at the bottom of the node. Type a question or prompt, choose a model from the dropdown in the top-right corner of the node (Claude, GPT-5, Gemini, or an open-source alternative), and press Enter or click Send.The model’s reply appears inside the same node, directly below your message. This is your first conversation thread.
You can change the model on any individual message, not just at the start of a thread. Use the model picker that appears next to each reply to switch mid-conversation.
3

Spawn a second chat thread

With your first node active, click New chat in the canvas toolbar (or press N). A second node appears on the canvas next to your first one. You can drag it anywhere to arrange your workspace.Type a related — or completely different — question in the new node and send it. Both threads are now live on the canvas at the same time. Resize, reposition, or zoom out to see them together.
There’s no limit to how many chat nodes you can have on a canvas. Threads you’re not actively reading stay on the canvas so you can come back to them at any point.
4

Fork a message to branch the conversation

Hover over any message bubble inside a node. A Fork button appears on the right side of the message. Click it.NodePad creates a new node that starts from that exact message — same conversation history up to that point, ready for you to take in a different direction. The original node is unchanged. Both branches run from the same starting context, so you can compare two different prompts or two different models on the same setup.
Forking is most useful when you want to test two different follow-up questions without losing either answer. Try asking the same question two ways and see which response works better for your use case.
5

Add a sticky note to lock in context

Right-click any node and select Add sticky note, or use the + button that appears when you hover over a node’s edge. Type a constraint or reminder — for example, “Always respond in bullet points” or “Assume the audience is non-technical.”The sticky note stays attached to the node. The model reads it as persistent context for every message in that thread, so you don’t have to repeat yourself.

​
What to try next

Forking and merging

Learn how to combine two branches back into one thread once you’ve found the direction you want.

Per-message model selection

Switch models mid-thread to get the best answer for each step of a complex task.

Sticky notes

Use sticky notes to enforce tone, format, or domain constraints across a whole thread.

Canvas basics

Learn keyboard shortcuts, zoom controls, and how to organize a busy canvas.