Skip to content
Auteryn

Web Search

Web Search

Section titled “Web Search”

Web search is a built-in capability that lets your agent look things up on the live internet — so its answers reflect current information instead of only what the model was trained on.

What it’s for

Section titled “What it’s for”

Current information

Recent news, prices, release notes, documentation, and anything that changes over time.

Grounded answers

The agent cites what it found, so you can trace an answer back to its source.

Research tasks

Gather, compare, and synthesize information across multiple pages in a single run.

Fact-checking

Verify a claim against live sources rather than relying on memory alone.

How it works

Section titled “How it works”
  1. The agent decides a task needs fresh information and issues a search query.
  2. It receives a set of relevant results from the web.
  3. It reads the most relevant pages to pull out the details it needs.
  4. It synthesizes an answer, referencing the sources it used.

You don’t have to tell the agent how to search — describe what you want to know, and it forms the queries itself. For long research tasks, run the agent in Deep mode so it has the time and steps to read across several sources.

Reading a specific page

Section titled “Reading a specific page”

Beyond open-ended search, an agent can read a specific URL you give it — pull the content of an article, a doc, or a public page directly into the task. Just paste the link into your message.

Tips

Section titled “Tips”
  • Ask for what you want to know, not the exact search terms — the agent phrases queries better than a keyword list.
  • Use Deep mode for multi-source research so the agent can read several pages before answering.
  • Ask it to cite sources when accuracy matters, so you can verify the result.
Section titled “Related”