Deep Research Agent

An AI-powered research assistant that automates comprehensive web research through a multi-agent pipeline. Given a research query, the system generates clarifying questions to focus scope, plans targeted web searches, executes them in parallel, and synthesizes findings into a structured 5-10 page markdown report. The pipeline features built-in quality evaluation loops — a Search Plan Evaluator and Report Evaluator score outputs on criteria like coverage, depth, and completeness, triggering automatic revisions if quality thresholds aren't met. The architecture uses 7 specialized agents (Clarifier, Planner, Search, Writer, and 3 Evaluators) orchestrated through OpenAI's Agents SDK, with a Gradio UI that progressively reveals each phase of the research process.

Project Gallery

❮❯