Capturing web pages

The Ultrathink extension makes it simple to save any web page to your knowledge base. You can capture full pages, screenshots, selected text, or even all your open tabs at once.

Basic capture

To save the current page:

1. Navigate to the page you want to capture
2. Click the Ultrathink extension icon in your toolbar (or press Ctrl+Shift+4 on Windows/Linux, Cmd+Shift+4 on Mac)
3. The capture popup appears with auto-extracted details:
   • Title: pulled from the page's <title> tag or <h1> heading
   • Type: auto-detected based on the URL (for example, "link", "video", or "chatgpt")
   • Notes: an empty field for your own thoughts
4. Add any notes you want to attach to the entry
5. Click Save

The entry is created immediately and sent through the AI processing pipeline for summarisation, topic extraction, and relationship mapping.

What gets extracted

When you capture a web page, Ultrathink automatically extracts the following metadata:

Field	Source
Title	Page <title> tag or first <h1>
URL	Current page address
Description	Meta description tag
Author	Meta author tag or detected byline
Published date	Article date, when available
Favicon	Site icon for visual identification
Content type	Auto-detected from URL pattern

You can edit any of these fields in the capture popup before saving, or later in the detail panel of the web app.

Screenshot capture

Capture a visual snapshot of the current page:

1. Press Ctrl+Shift+5 (Windows/Linux) or Cmd+Shift+5 (Mac)
2. Alternatively, click the camera icon in the capture popup
3. The screenshot is taken of the visible viewport and attached to your entry automatically

Screenshots are processed by AI to generate a visual description and extract any visible text via OCR.

Requirements: screenshot capture requires Chrome or Edge version 127 or later. Some pages (banking sites, DRM-protected content) block screenshot functionality due to browser security policies.

Selected text capture

Save a specific passage instead of the whole page:

1. Highlight the text you want to capture on any web page
2. Press Ctrl+Shift+4 (Windows/Linux) or Cmd+Shift+4 (Mac)
3. The capture popup opens with the selected text pre-filled
4. Add context notes if needed
5. Click Save

The entry is saved as a snippet type, which stores both the selected text and the source URL for reference. This is useful for saving key quotes, definitions, or passages without capturing the entire page.

Bulk tab capture

Save all your open tabs in one action:

1. Open the Ultrathink extension popup
2. Check the All tabs option
3. Click Save

Each tab becomes a separate entry in your knowledge base. If you use Chrome's tab groups, the group name and colour are preserved as metadata on each entry. This is particularly useful when you have a research session with many tabs open and want to save everything before closing them.

Auto-save

The extension automatically saves your capture so you do not need to click the Save button manually:

• After you stop interacting with the capture popup, a 3-second countdown begins
• A progress bar shows the countdown visually
• To cancel, click Cancel or start typing in the notes field to reset the timer

Content type auto-detection

The extension automatically identifies the type of content you are capturing based on URL patterns:

URL pattern	Detected type
chat.openai.com	ChatGPT conversation
claude.ai	Claude conversation
perplexity.ai	Perplexity search
youtube.com, vimeo.com	Video
linkedin.com/in/	Profile
notion.so	Notion page
SharePoint, OneDrive	Microsoft Office document
URLs ending in .pdf	PDF document

When the auto-detected type does not match what you want, you can change it manually in the capture popup or later in the web app's detail panel.

After capture

Once an entry is saved, it goes through the AI processing pipeline automatically:

1. Summary: AI generates a concise summary of the content
2. Classification: topics, people, and entity types are extracted
3. Relationships: related entries in your knowledge base are identified
4. Similarity: near-duplicate or closely related content is flagged

You can monitor processing progress in the entry's detail panel within the web app.

Tips for better captures

• Add notes at capture time: jot down why you are saving the page, key takeaways, or action items; this gives the AI more context for classification and makes the entry easier to find later
• Use topics consistently: if you have existing topics in your knowledge base, the AI will attempt to match captured entries to them; review and correct topic assignments to improve future accuracy
• Capture before the page changes: some content (social media posts, live dashboards) can change or disappear; capture promptly to preserve the current state
• Check the detected type: auto-detection works well for common URL patterns, but if a page is misclassified, changing the type ensures the right processing pipeline is used