Chat

The chat panel on Cortex Canvas gives you two modes: standard chat and CASi. Both live in the same interface, but they draw from different levels of context.

Standard chat

Standard chat reads your prompt and returns a direct answer. It has no awareness of your canvas nodes, brand context, or competitor data. Use it for quick tasks: summarising a paragraph, answering a factual question, or drafting something that doesn't need your brand voice.

You can switch between different AI models within the same conversation.

CASi chat

CASi reads your full workspace. Brand terms, client notes, prior articles, research nodes, and source material all stay available within the thread. When CASi references something, it can point to the specific node behind the claim.

Use CASi when the output needs to carry your brand voice, reference your research, or draw from competitor intelligence. CASi is the right choice for drafting content, refining arguments, and building on what's already on your canvas.

Features available in both modes

• Model selection. Switch between different AI models mid-conversation. The task gets matched to the right model without copy-paste.
• Web search. Capture live sources in-thread. Current claims stay testable.
• Memory. Session history is stored on the canvas. The next conversation starts where the last one stopped.
• Brand context. With CASi, your workspace documents are connected and product language carries forward automatically.

Using chat alongside the canvas

Chat and the canvas work together. You can ask CASi about a specific node, request a summary of your research, or have it draft a section based on selected nodes. The chat output can then be saved as a new node on the canvas, feeding back into your research base.