Learn GPTree

Getting started

• Getting started with GPTree

  What GPTree is, how Branch, Compare, and Merge fit together, and exactly what to do in your first five minutes. Free to start, no card required.

• Find your way around GPTree

  A tour of the screen you sign in to. What Home shows, what the left rail and the panel beside it do, where Settings and Help live, and how the same layout folds onto a phone.

• Your first branch: the moment GPTree clicks

  The first time you branch a conversation is when GPTree makes sense. What happens when you click Branch, what it feels like, and how to get there.

• Import your ChatGPT history into GPTree

  Install the Chrome extension, open a ChatGPT conversation, and one click brings it into GPTree ready to branch. Setup takes about two minutes.

• Set up your first project

  Projects are folders for your AI work. They group related conversations and hold instructions and files the AI reads. Here's how to set one up well.

• Using GPTree on your phone

  The mobile app is the same GPTree you use on the web, with your conversations, branches, and projects on your phone and notifications that deep link.

• Getting help inside GPTree

  Where to find help in GPTree, from the Onboarding Guide to in-chat answers with deep links, guided tours, the checklist, the Discord community, and human support by plan.

Features explained

• What branching does, and why linear chat falls apart

  Branching forks a conversation at any useful point so you can try a new direction while the original thread, and all its context, stays intact.

• Comparing branches side by side

  Put two branches of one conversation next to each other and read a diff that shows exactly where the answers agree, differ, and which one to keep.

• Where your branches disagree

  The Agreement and conflict panel shows what your branches do not agree on, including the branch that qualified a claim everyone else stated flatly.

• What merging means in GPTree

  Merging brings the useful outcome of a branch back into your main flow, from picking the better answer to synthesizing several outputs into one.

• The Merge Board: turn many branches into one answer

  A full-page workspace that turns several branches into one final output. Pick a summary, a decision with pros and cons, or an export-ready draft.

• Explore mode: one prompt, up to five answers

  Send one prompt as two to five side-by-side variants, each with its own model. GPTree scores the answers automatically, then you pick a winner or merge the best of all of them into one answer.

• How projects keep your AI work organized

  Projects nest inside projects, so your AI work gets real folders, plus instructions, reference files, and scoped memory that brief every new chat.

• How GPTree memory works

  GPTree saves facts, decisions, and patterns from your chats, scoped to a conversation, project, or account. Here's how it works and how to edit it.

• System personas: six ready-made ways to be answered

  Every GPTree account comes with six personas built for the work people actually bring: engineering, writing, studying, opening a decision up, closing one down, and content. Pick one per message, set one as your default, or switch off the ones you don't want.

• What agents are in GPTree

  GPTree agents form a company: an org chart, delegation, crews, skills and routines, a decisions ledger, and approval gates with a real audit trail.

• Connecting a knowledge source

  On a paid plan, GPTree can answer from an outside knowledge base and cite the exact passage it used. The first built-in source is the Scryfall Magic: The Gathering card database.

• Bring your own API keys

  Add your own OpenAI, Anthropic, Google AI, or Mistral keys and their models appear in the picker. GPTree organizes the work; your provider bills usage.

• The desktop app and Chrome extension

  The Chrome extension pulls ChatGPT conversations into GPTree for branching. The desktop app adds Watching, opt-in context capture that runs locally.

• Document citations: answers that show their source

  When a chat answer draws on one of your project reference documents or a file you attached, GPTree adds a small numbered citation chip after the claim. Click it to see the exact passage the answer came from, as the document said it at the time.

• The Knowledge API: your own documents, cited, in one call

  Pull cited passages from your own documents and conversations into any tool you already use. One call returns a ready-to-use block of evidence with citations attached, and you choose whether GPTree writes the answer or your own model does.

• Knowledge: one place for every document

  The Knowledge page collects every document GPTree knows about, whatever produced it, and keeps each one's history so you can read the current version, read an older one, and see exactly what changed.

• Find a document by what it says, and see where it was used

  Knowledge searches inside your documents, not just their titles, and every document shows which of your answers cited it.

• Point your own AI at your documents

  GPTree can act as a knowledge server that your own AI assistant connects to, so the assistant you already use can search your documents, quote them exactly, and tell you when it cannot answer.

• Ask what your documents said at a point in time

  The Knowledge API can answer from the corpus as it stands now, from a corpus pinned at the start of your run, or from the corpus exactly as it was when an earlier answer was given. You name which one, and every response says what it actually guaranteed.

• What GPTree Image Studio is

  An editor that puts real text and image layers on top of a generated picture, saves each state as its own branch, and only spends credits when you ask the model for a change.

• How GPTree searches the web

  When a question needs current information, the assistant runs a live web search itself and answers from the results, naming its sources. You can also force a search for one message with the Web Search button. It will never describe a search it did not actually run.

• Keep decisions and merges in your knowledge

  Switch keeping on for a project and the work you finish becomes something GPTree can cite later: a merge you applied, a decision you recorded, a board you committed. It lands in your own space, readable by nobody else until you share it.

How-to guides

• Ask the assistant to read and compare your branches

  In any conversation, ask the assistant to look at your branches and it can read them, compare the answers, and recommend one, right in the chat.

• Finding things with search

  How search works across your projects, conversations and messages, what the result counts mean, and what to do when search says it is temporarily unavailable.

• Make and manage your personas

  A walk through the Personas page. Choose which persona new chats start with, create your own from a blank form or a starting point, and switch built-in personas off so the picker only holds the ones you use.

• Start a conversation your way

  The New conversation button has a menu behind it. Start from a document, open a blank document, hand a task to an agent, or make a new project, from Home or from the left rail on any screen.

• How to merge a branch back

  Pick the messages worth keeping, choose one of four targets and four strategies, check the Preview Merge diff, then Confirm Merge to make it real.

• Manage your projects

  The Projects page lists every project as a nested tree with counts and last-opened dates. Create, pin, open and tidy from there; each project's settings page holds its instructions, files, defaults and sharing.

• How to share a conversation or project

  Create a read-only public link anyone can open, or invite teammates as viewers or editors on a paid plan, and know where shared work shows up.

• How to attach files to a chat

  Attach PDFs, Word and Excel files, images, and source code to any message. GPTree extracts the text in the background so the model can read it, quote it, and answer follow-ups about it.

• How to see where an answer came from

  Click the numbered chip after a claim to open the exact passage in your document that backs it. A quick guide to reading, trusting, and troubleshooting citation chips.

• How to create a Knowledge API token

  Create, copy, and delete the tokens that let your own tools pull cited context from your GPTree documents and conversations, and choose whether a token can also spend credits on generated answers.

• Collections: keep the documents for a job together

  A collection is a named set of documents you keep together for a job, so you can find them in one place without making copies.

• How to keep your own documents in sync

  Send documents from SharePoint, Google Drive, a network share, or your own machine into GPTree from your own code, retry safely, remove what your source removed, and hand over files too large to process on the spot.

• How to send a coding task to Cursor from a conversation

  Work out a change in a conversation, then hand the coding task to Cursor's cloud agents from the same place. Add your Cursor API key once, pick a Cursor model, and send the task with your GitHub repository link. The reply shows the Cursor job and, when it finishes, the result.

• How to create an agent

  Set up an agent with a name, a role, and a persona, then chat with it in its own pane like a teammate. Its questions and approvals land in your inbox.

• How to verify your email

  New accounts confirm their email before running the AI. Click the link we send, or use the Resend button in the banner. Google and Apple sign-ins are already verified.

• How to correct what GPTree assumes about you

  GPTree carries a short profile and a set of memories between conversations. If either one describes your work wrongly, here is how to fix it so the correction sticks.

• How to add text and layers to an image

  Open Image Studio on a generated picture, type editable text straight onto the artwork, style it, and save it as a branch. Layer saves are instant and use no image credits.

• How to talk with GPTree by voice

  Start a voice session and just talk. GPTree replies straight away, keeps listening while your AI model writes the answer, then gives you a short spoken summary. With Live voice it answers as you talk and you can cut in at any time. Say "read it" for the whole answer. You can also dictate a message or hold the mic for a single command. Step by step for the browser, the Mac app and the phone apps.

• Voice, your microphone and your privacy

  When GPTree listens (including in a voice session), where your speech is turned into text, what Live voice sends to OpenAI, what GPTree keeps (your words, never recordings), how spoken replies are made, how long your words are kept, and how to see, delete or stop keeping them.

Ways people use GPTree

• Get three models' feedback on a spec, merged into one review

  Paste a plan or spec from your coding agent, have three models review it side by side, and send back one merged piece of feedback instead of three tabs of opinions.

• Debugging with branches: try three fixes without losing the thread

  Branch a debugging thread at the last good message, run three candidate fixes in separate paths, and keep the context you spent all morning building.

• Finding the right voice by branching the draft

  Highlight the paragraph that almost works, branch it into formal, casual, and witty versions side by side, and hear which one actually sounds like you.

• Headlines that stay readable on any photograph

  Type set over a picture fails on contrast, not on wording. Image Studio gives you the shadow, outline and reflow controls to fix that, and a one-press treatment that sets them for you.

Plans, billing, and policies

• The GPTree tree pledge: a tree a month for every paying customer, and one for every referral

  Every month, for every person paying for GPTree, we fund one real tree through Greenspark, plus an extra tree for every referral who subscribes. Here is what that means, why it is monthly, and what we deliberately do not claim.

Developer documentation

Calling GPTree from your own tools: endpoint reference, a sync client, worked examples, and MCP setup.

• Knowledge API reference

  Endpoints, request fields, response shapes, status codes and rate limits for /api/kee/v1.

• Integration guide

  The shape of a working sync: delivery identity, explicit deletion, background jobs, and reading the judgment on an answer.

• Worked example

  A full integration transcript with real requests and responses, including a refusal.