An AI workspace: every model, one canvas
Most AI work does not fit one chat. A real problem turns into five conversations, in two apps, with no memory of each other, and you become the courier pasting context between them. An AI workspace fixes that: one place where every conversation about a problem lives together, visible at once, and any model can be pointed at any part of it.
fork
When a side question threatens to derail a chat, fork it onto its own path from that exact point. Both paths keep the full history behind them; neither pollutes the other.
reference
Pull an exact earlier message into scope with @, whichever chat it lives in. No re-explaining, no pasting walls of text and hoping the model keeps it straight.
merge
When separate chats each hold a piece of the answer, merge them into scope and ask for the synthesis. One draft, written with everything you chose in view.
One workspace per problem, not per app
Open a workspace for the problem, not the tool. Ask your first question the way you would anywhere; the difference is that the reply lands on a canvas with room around it, not at the bottom of a scroll you will lose.
Let the tangents be tangents
The moment a chat wants to go two directions, fork. The main thread continues clean, the tangent gets its own path, and both stay on screen. Nothing gets abandoned in a scrollback; nothing poisons the conversation you cared about.
Pick the model per message
The model is chosen per message, not per workspace. Draft with a fast model, then re-ask the load-bearing question with a frontier one. Every model sees the same context you assembled: nothing hidden, nothing injected behind your back.
End with a synthesis, not an archive
When the exploring is done, merge the branches that mattered and ask for the final answer. The workspace is not a pile of dead chats; it is the map of how you got to the conclusion, still legible next week.
Common questions
What is an AI workspace?
A place where all your AI conversations about one problem live together and can be worked as a whole: branched, cross-referenced, and merged, instead of isolated linear chats in separate tabs.
How is a workspace different from a folder of chats?
Folders organize chats; a workspace connects them. In NodePad the chats themselves sit on one canvas, so you can see every thread at once, reference a message from one chat inside another, and merge several into one answer. A folder cannot do any of that.
Can I use different AI models in one workspace?
Yes. Claude, GPT, and Gemini work in the same workspace and the model is picked per message, so different models can answer different parts of the same problem with the same context in view.
Does the model know about the workspace?
No. The model receives exactly the context you assembled, the path behind your message plus anything you referenced or merged, and nothing else. No hidden system prompts, no injected summaries. That control is the point.
No credit card to start. Your first workspace is already waiting.