Research & Compile AI Agents: 11 Real Examples

This directory currently includes 11 recipes for the research & compile pattern, across 7 roles. Research and compile agents turn a question into a focused report built from multiple sources. They can search, collect evidence, compare findings, and organize the result around a decision or task. This pattern is useful when the challenge is not finding one answer, but assembling a trustworthy view from scattered material. The recipes cover different roles and topics, yet the useful design principles remain consistent: define the question, select acceptable sources, cite what was used, and show uncertainty rather than smoothing it away. A research agent should help someone get to the evidence faster, not create a polished answer with no traceable basis. Start with a recurring research task where the final output has a clear audience, format, and decision owner. Every example makes its inputs, output, and handoff concrete, so you can compare real workflow boundaries before adapting the pattern to your own tools and process.