How GPTree organizes your imported history into folders

A ChatGPT export is one long undifferentiated list. Several hundred conversations, no structure, titles that were auto-generated and never looked at again. Dropping that into GPTree unchanged would just move the mess.

So the import has an Organize step. GPTree proposes a folder structure, shows you exactly what is going where, and lets you change any of it before a single project is created.

How the proposal is built

Two sources, in order of authority.

First, whatever your export already knows. If a conversation belonged to a ChatGPT Project, or ran inside a custom GPT, that is grouping you set up yourself, and GPTree uses it rather than inventing a new one. Those folders keep the names you gave them.

Second, for everything else, GPTree groups conversations by what they are about. The first pass runs in your browser on titles and dates alone, which is why the Organize step appears immediately rather than after a wait. Once the upload starts, a closer pass reads a short digest of each conversation and regroups anything you left on automatic.

What you can change

• Rename a folder.
• Dissolve a folder you do not want. Its conversations become unsorted.
• Move an individual conversation to a different folder, or out to unsorted.
• Apply bulk rules, like dissolving folders with only a couple of conversations, or leaving very short conversations unsorted.
• Change the name of the container everything goes into.

The rule about your changes

Anything you touch is kept exactly as you set it. The refinement pass that runs during upload only reconsiders folders and conversations you left alone.

This matters because the alternative is worse than no organization at all. A tool that quietly re-sorts your decisions after you have made them is a tool you stop trusting, and then you stop using the feature.

If you would rather not think about any of it, click through. The proposal is already accepted by default.

Where the folders end up

Folders are created as projects nested inside one container project, named after where the history came from ("ChatGPT history" or "Claude history", for example, or "AI chat history" when an import mixes sources) unless you change it.

That container exists so importing does not scatter twenty new projects through a list you have already organized. Everything from the import is in one place, easy to review, and easy to remove in one action if you want to start over. Once you are happy with a folder you can move it out to the top level like any other project.

Conversations that did not fit a folder are not filed anywhere new: they stay where the import put them, the project you imported into or outside any project, rather than in a folder called "Miscellaneous". An unsorted conversation is more findable than one filed under a name that means nothing.

When folders are skipped

If you are importing only a handful of conversations, GPTree does not create folders at all. Structure for eight conversations is ceremony, not organization. The threshold exists so a small first import stays simple.

Importing again

Re-importing refills the same folders rather than creating a second set beside them. Folders are matched by name inside the container, so the structure you settled on is the structure you keep.

Recovered branches follow their conversation

If the import recovered branches from regenerated answers, those branches go into the same folder as the conversation they came from. A conversation and its branches are never split across two places.