> ## Documentation Index > Fetch the complete documentation index at: https://docs.node-pad.com/llms.txt > Use this file to discover all available pages before exploring further. # Research > Keep separate investigations visible on one canvas and synthesize them only when you are ready. export const EdgeChip = ({type = 'continuation', children}) => {children}; Research gets messy when every sub-question lives in the same transcript. NodePad works better when you split the work into clean branches, keep those branches visible, and pull the useful pieces together deliberately — usually through a merge at the end. ## A pattern that works Start with a framing branch that defines the scope of the question and names the sub-questions you need to answer. Create one branch per sub-question so each line of inquiry has its own context and output history. When one branch uncovers a source that matters elsewhere, use an `@` reference instead of summarizing it by hand. Once the branches have enough signal, merge the important messages into a new draft and ask for a comparison or summary. ## Why it works Each branch stays focused on one topic instead of inheriting every tangent. You can see where a conclusion came from because the source branches stay on the canvas. The final summary draws from real source branches instead of from your memory of what happened in them. Add a text note naming the research question and the standard for evidence you want to use. It helps when the workspace grows past a few branches. ## Continue Compose a synthesis draft from selected source messages. Pull a single message back into another branch with `@`.