Using Web Search
Give your content real-time depth: Web Search pulls in current data, statistics, and sources automatically.
What It Does#
When Web Search is enabled, the Content Agent searches the internet for relevant, up-to-date information related to your prompt. It decides on its own when fresh data would improve your content, no manual triggering required.
During research, a Web Search tool indicator appears in the chat with a spinning icon while working, then a checkmark when results are found. The agent cites discovered sources inline in its response.
Default Setting#
Web Search is enabled by default for podcast episodes. The agent proactively finds current data, statistics, and articles to support your topic.
How to Toggle Web Search#
- Click the Settings button (gear icon) in the chat input controls.
- Find the Web Search checkbox at the top of the dropdown.
- Check or uncheck "Use External Web Research" to toggle it on or off.
The setting takes effect immediately for the current conversation.
When to Keep It On#
- Your topic involves current events, trends, or recent developments
- You want supporting data and statistics the agent can find for you
- You're covering a broad topic and want diverse perspectives
- You want citations and references in your content
When to Turn It Off#
- You only want the agent to use your attached sources
- Your content is based entirely on proprietary or internal material
- You're working with sensitive topics where you want full control over information used
- You've already provided all the source material the agent needs
You can attach your own sources and keep Web Search on at the same time. The agent weaves both your materials and its web research into a cohesive piece.
Frequently Asked Questions#
Does web search use my credits?
No. Web search and all research activities are completely free. Credits are only consumed when you publish or download content.
Can the agent access content behind paywalls?
No. The agent can only access publicly available web content. If you need the agent to reference paywalled content, download or copy the text and attach it as a source directly.