Release Notes: Source-Grounded Chat, Multi-Link Imports, and Live Document Progress

DocTree can now bring more learning material directly into the canvas and use it as reliable context for conversation. This release adds automatic canvas context, source-grounded chat, multi-link imports, dedicated document items, and clearer processing progress.

Highlights

• Select a canvas item to automatically add it to the chat context.
• Ask questions about selected items or request changes to editable items.
• Ground answers in selected PDFs, documents, and crawled articles.
• Import multiple articles or YouTube links at once.
• Follow document upload and processing progress in real time.
• Move useful chat results and recommendations onto the canvas.

What is new

Select an item, then ask about it

Selecting an item on the canvas now automatically adds it to the chat context. You can ask DocTree to explain it, summarize it, compare it with other selected material, or use it as the basis for your next request.

For editable items—including notes, summaries, articles, code, curricula, resource cards, videos, books, and people—you can ask for a change in plain language. DocTree updates the selected item directly, keeping the conversation and canvas connected.

Answers grounded in your selected sources

You can select an uploaded PDF or document on the canvas and ask questions based on its indexed contents. Crawled articles work the same way: select an article and DocTree uses the extracted source material when answering.

This gives you more control over what informs a response. Instead of depending on general AI knowledge, you can explicitly choose the materials DocTree should use. When a selected source is still processing or could not be indexed, the chat provides clear feedback rather than answering without the expected context.

Import several learning sources at once

You can paste multiple article or YouTube links into chat and import them together. DocTree removes duplicate links, processes supported sources in parallel, and reports individual failures without discarding successful imports.

Imported articles become visible canvas items connected to their original pages. YouTube links are represented using the appropriate video experience. This makes it faster to assemble a workspace from reading lists, research tabs, tutorials, and reference collections.

A more complete document workflow

Uploaded PDFs, DOCX files, and text files now appear as dedicated document items on the canvas. You can keep source material beside the articles, notes, and generated content that depend on it.

Document previews open inside DocTree, with support for reading PDFs and extracted text without switching applications. The original file remains available to download.

Uploads now show separate upload and processing progress, providing clearer feedback during longer imports and preventing questions from being submitted before their source material is ready.

Turn useful chat results into canvas material

More kinds of chat output can become reusable canvas items. Code blocks can be moved into dedicated code cards, while individual videos, articles, books, and people can be saved from resource recommendations.

You can also select excerpts from earlier chat messages and include them in a new request. The composer shows how many excerpts and canvas items are attached, making follow-up questions more precise.

More focused resource recommendations

Resource requests now distinguish between videos, articles, books, and people. When a request is too broad, DocTree can ask which material would be most useful.

Recommendation handling also filters repeated results more carefully, producing a cleaner reference shelf across related requests.

Why it matters

Learning rarely starts from one perfectly prepared source. It begins with browser tabs, documents, videos, code, earlier conversations, and references discovered along the way.

This release connects those materials more directly. Selecting something on the canvas is now enough to continue the conversation around it. You can investigate a source, ask for an explanation, or reshape an editable item without rebuilding the context manually.

Smoother experience

Document progress is retained more reliably, selected chat text behaves more consistently, and partial link-import failures provide clearer feedback. Additional safeguards around article fetching and source selection help keep grounded answers dependable.

Together, these changes make DocTree a more responsive learning workspace—one where your sources, questions, and canvas remain connected as your understanding develops.