How to build a moodboard from posts you have saved
Most people who save things online have a moodboard already. It is spread across likes, bookmarks, screenshots and open tabs. The work is not collecting more. It is choosing.
Start with one question
A moodboard answers something: what should this room feel like, which direction for this poster, how formal should the onboarding be. Write the question at the top before you pull anything in. It decides what stays.
Gather from where it already is
Pull references from every place you keep them: the posts you liked, the screenshots on your phone, links you meant to read. Do not judge yet. Put them in one place so you can see them side by side, because the pattern only shows when they sit together.
Group by what you notice, not by source
Sort by what the references share: a colour, a kind of light, a type of lettering, a spacing habit. Name each group with a plain phrase, such as “dark and quiet” or “thick outlines”. If a reference fits no group, it is probably not for this board.
Cut until it says one thing
Keep the references that answer the question and remove the rest, even good ones. Twelve that agree say more than forty that do not. The cuts are not lost. They stay in your collection for a different question.
Write what to borrow
Next to each reference, note the one thing you want from it: the spacing, the palette, the way the heading sits. This is what makes a moodboard usable by someone else, and by you in a month.
Doing it in Curatively
Keep a vault for the project so its references stay apart from everything else. Saves from the share sheet, the camera, a pasted link or an import land in the vault, and with AI processing on, each gets a description, tags and colours. You can then browse by colour or search for “warm kitchen with green tile” in your own words.
You can also hand the vault to Claude, ChatGPT or Cursor over MCP and ask it to work from your references. See giving your AI assistant your taste.