Working with Results

This page follows your material through the app: what you put in, what you take out, and the shortcuts that make both faster.

Attaching files

You can upload files and use them as context in chats.

Supported formats

• Images: PNG, JPG, GIF, WebP
• Documents: PDF (text extraction)
• Text: MD, TXT, CSV, JSON, code files
• Office: DOCX, XLSX (text and table extraction)

Limits

• Max 10 MB per file
• Max 5 files per message (files and research documents combined)
• Of these, at most 3 research documents per message

Upload via the paperclip in the composer or Cmd+U. Attachments show as a preview: thumbnails for images, badges for documents.

How a file becomes context

On upload, the content of each file is extracted server-side: text and code are read directly, while text and tables are pulled out of PDF, DOCX, and XLSX. When you attach a file to a message, that extracted text flows into the chat response as context.

Images are the exception. They are not extracted as text but passed to a vision-capable model as an image (see Vision).

Two routes lead to the same place:

• Project files are assigned to a project, appear in the sidebar, and can be attached again and again in that project's chats.
• Directly attached files are added to a single message as you write it.

In both cases, a file only becomes context once it is attached to the message in question. Files and research documents can be attached together (see Research): files provide raw material, research documents provide already-prepared, reusable knowledge from earlier runs.

Attached file context applies to chat responses. Agents and research do not use it directly; their results are stored as research documents that you can later attach in a chat.

What matters: attach the relevant files specifically, not entire archives. Very long extracted text is truncated; a few focused files give better context than a large, unspecific pile.

Saving responses as a file

Responses can be saved as a document or a file. The app groups both under Artifacts.

Documents and files

• Documents (automatic): when the system recognizes a request for a standalone, reusable result, the matching response can be saved as a document automatically. This can include documents and suitable tables in particular. You do not have to explicitly use the word "artifact". Not every long response and not every table is saved.
• Files (manual): via Save as file below a response, available at any time. Automatic saving does not replace this manual path. If a response contains a chart, the chart is also saved automatically as a file.

When automatic saving succeeds, the chat may show the response in a compact form with a button to open the saved document; the full content then need not be repeated in full in the chat as well.

To save a response manually as a file:

1. Below an assistant or agent response, choose Save as file.
2. Select the desired format: Markdown (.md), CSV (.csv), Excel (.xlsx), PDF (.pdf), or Word (.docx); for responses with a chart, also Chart (.svg).
3. The artifact is generated and stored server-side.

For CSV and Excel, tables from the response (Markdown tables) are carried over; otherwise the text is stored line by line.

After manual saving, an artifact chip with a download appears below the response. The sidebar has an Artifacts section, per project or under "Recent".

Creating artifacts requires an active subscription.

The artifact overview

Artifacts opens a shared overview of your documents and files. The two kinds remain separate entries: documents are labeled Document, files show their formats.

• In a desktop browser, the overview shows a grid with previews by default; the toolbar switches to the list view. There you can search by title, filter by All, Documents, or Files, and sort by Recently updated, Date created, or Alphabetical.
• On mobile devices and narrow screens, the overview appears as a list, without this toolbar.

Opening, copying, download, and export

Documents open in the document view, where you can copy the content. Downloading or exporting documents to other formats is not currently available.

Files open differently depending on the format:

• Markdown, CSV, and Chart open in the view. There you can copy the content and save the file as Markdown, CSV, Excel, PDF, or Word from the menu; Markdown files also offer a dedicated button for Word. These versions are generated again from the original response and are only available as long as that response still exists.
• PDF, Word, and Excel are downloaded in the browser.

Deleting

• Files can be deleted individually per format in the overview and in the sidebar (file and entry).
• Documents have no delete action of their own in the overview. Deleting the associated chat removes the documents tied to that chat; deleting your account removes all saved documents (see Privacy).

Edited responses

An automatically saved document keeps the version in which it was saved. If you later edit the associated response in the chat, the document does not change; there is no automatic synchronization. If the response and the document differ, the chat shows the full response and the document card as well.

Keyboard shortcuts

Keyboard shortcuts are a web feature. On macOS use Cmd, on Windows and Linux use Ctrl.

Action	Mac	Windows
New chat	Cmd+K	Ctrl+K
Send message	Cmd+Enter or Enter	Ctrl+Enter or Enter
New line	Shift+Enter	Shift+Enter
Toggle Voice	Cmd+Shift+V	Ctrl+Shift+V
Upload file	Cmd+U	Ctrl+U
Open settings	Cmd+,	Ctrl+,
Push-to-talk (recording)	hold Shift	hold Shift
Close modal/dialog	Escape	Escape

You can also find an overview in the settings under "Keyboard Shortcuts".

Security

The same rules cover uploaded files and generated artifacts:

• Uploaded files are validated via MIME type and magic bytes, which protects against misdeclared files.
• Storage is in private object storage; delivery runs through short-lived, signed URLs.
• The content of an artifact is loaded server-side from your own message or your own run. The client does not send any file content.
• Access is restricted to your own account (Row-Level Security).