An AI agent that searches the web and writes briefs with sources
The Research Agent step searches the web and synthesizes findings using AI. It supports two actions: Search produces a narrative summary with citations, and Structured Search populates user-defined fields from the research results.
Search the web and produce a markdown summary with cited sources. Write a research goal as a prompt, and the agent issues multiple web searches, reads the results, and synthesizes a brief.
Configure a provider, model, and search depth (Quick runs up to 2 searches, Standard up to 5, Deep up to 10). Write a prompt with references to data from previous steps. Advanced options let you restrict or exclude specific domains and filter by recency. Output fields are pre-configured: a summary in markdown, a list of cited sources (with title, URL, snippet, and published date), the number of sources, and the number of searches.
Search the web and populate user-defined structured fields. Define custom output fields like employee_count, headquarters, or recent_funding, and the agent fills them from its research instead of writing a narrative summary.
Same search settings as the Search action (provider, model, depth, domain filters, recency). Define your output fields to specify what data you want extracted. Citation data (sources, number of sources, and number of searches) is included automatically alongside your custom fields.