Run web searches and return ranked results

Brave Search API returns a set of ranked web results, including titles, snippets, and links, for a given query. A Beam agent runs a search when a trigger fires, such as a request to check a fact, price, or current event, and reads the results returned. It applies the approved rule for what to do next, such as pulling the top result into a report or comparing it against existing data. Where the results are ambiguous, conflicting, or touch a sensitive topic, the agent surfaces the raw results to a person rather than picking an answer on its own.

Pull fresh facts into an agent workflow

Because Brave Search draws on an independent web index, a Beam agent can use it to check whether information already stored elsewhere is still current, such as a company detail, a public price, or a status. On a schedule or trigger, the agent submits a query, reads the returned snippets and links, and applies an approved rule, for example flagging a mismatch between the stored value and the fresh search result. Any update to a record based on what it finds is queued for a person to confirm before anything is changed.

Filter and store results for later reference

Search results returned by the API can be saved so a workflow does not have to re-run the same query repeatedly. A Beam agent reads new results, filters them against approved criteria such as domain or recency, and stores the relevant ones alongside the record that triggered the search, for example a lead or a research task. Results that fall outside the filter, or where no clear match is found, are left for a person to review and decide whether a broader or different query is needed.