Search the web
Work can browse the internet in real time and read specific web pages as part of a task. Web search answers questions with up-to-date, sourced information. Open URL lets Work read and analyze a page you already have the link to.
Knowledge cutoff: every language model is trained on data collected up to a fixed date. After that date, the model has no built-in knowledge of new events, releases, or facts. Web search and Open URL let Work go past that cutoff and answer with verifiable, up-to-date sources.
For multi-source, in-depth research with structured cited reports and PDF export, use Deep Research in Chat. Deep Research isn't available in Work.
Web search
Web search lets Work fetch current information from across the web and use it to answer your prompt.
Activation
- Click the
+icon or type/in the chat window. - Select
Toolsthen enableWeb search.
When you ask a question that needs current information, Work searches the web and weaves the findings into its response. Ask your question naturally.
Reading the results
When Work uses web search to answer your question, you'll notice:
- A globe icon next to the response, confirming that web data was used.
- Inline links pointing back to the original sources.
- A
Sourcesbutton at the bottom of the response that opens a panel listing every reference in one place.
Every web-sourced response comes with its references so you can check the underlying material yourself.
News search
For news queries, Work draws from professional news partners Agence France-Presse (AFP) and Associated Press (AP), so results come from verified, editorially vetted sources you can cite with confidence.
When news results are included, you'll see a news icon alongside the globe icon. Sources link directly to the original agency reporting, giving you reliable material for briefings or internal updates.
For quick competitive or market intelligence, try prompts like "Latest regulatory updates in [your industry]" or "Recent funding rounds in [sector]."
Open URL
Open URL lets you bring web content directly into a Work chat. Instead of copying and pasting text from a webpage, paste the URL and let Work fetch, read, and use the page content as context.
How to use
- Ensure Web Search is enabled:
- Click the
+icon or type/in the chat window. - Select
Tools. - Enable
Web Search. Without this, Work cannot browse the web.
- Click the
- Paste a URL into the message box, along with your question or instruction.
- Send the message. Work fetches the page and uses its content to answer.
You'll see a link icon and an Opened Page mention in the response, confirming that Work read the page content.
Open URL also works with links to online files (PDFs, documents). Work fetches the file and answers your question with context.
Working with multiple URLs
You can paste several URLs in a single chat. Work keeps the content from each page in context, so you can ask follow-up questions that reference or compare them.
For example, paste two competitor product pages and ask: "Compare the feature sets described on these two pages." Work uses both pages as context to generate a side-by-side comparison.
Common use cases
- Competitor analysis: paste a competitor's product page and ask Work to extract features, pricing, or positioning.
- Documentation review: share a docs page and ask for a summary, or check it against your own specifications.
- Article summarization: drop in a long article and get the key takeaways.
- Meeting prep: paste an agenda or briefing document hosted online and ask Work to highlight the most important items.
Choosing the right tool
Pick the option that fits your situation:
| You want to... | Use |
|---|---|
| Get a quick, up-to-date answer with sources | Web search |
| Analyze a specific page you already have the link to | Open URL |
| Get a structured, cited report pulling from many sources | Deep Research (Chat legacy only) |
| Search trusted internal documents | Libraries |
| Pull live data from a connected tool (Gmail, Drive, Notion...) | Connectors |
Limitations
- Open URL fetches only the single page at the URL you provide. It doesn't crawl the entire site or follow links.
- Pages behind a login or paywall can't be accessed. Download the content and upload it as a file instead.
- Some highly interactive websites may not load fully, which can result in incomplete content.
- Web search can return outdated or incomplete content depending on what the source page exposes.