Live Web Search

Linkup gives AI applications access to current web results through a search endpoint. A Beam agent queries Linkup when a task needs fresh information, reads the returned results, and applies rules your team set about which sources to trust and how recent a result must be. It passes vetted findings into the workflow and cites where each came from. When results conflict, look unreliable, or touch a sensitive decision, the agent hands the question to a person with the sources gathered rather than acting on shaky ground by itself.

Sourced Answers

Linkup can return answers grounded in web content with the supporting sources attached. A Beam agent requests a sourced answer for a question inside a larger task, reads both the answer and its citations, and checks them against your acceptance rules: requiring a minimum number of sources or a recent publication date. It records the answer and links for an audit trail. Anything that would drive a customer-facing claim, a price, or a commitment is routed to a human to confirm before it is used anywhere.

Real-Time Retrieval

Linkup retrieves web content on demand so an agent works from current pages rather than stale training data. A Beam agent pulls the latest details for a lookup, reads the fetched content, and updates the record it is working on under your rules: refreshing a company detail or noting when a page has changed since last checked. It timestamps what it retrieved. Judgments about acting on ambiguous or fast-moving information stay with a person, whom the agent notifies with the retrieved material and its source.