Perplexity-Style Iterative AI Research with Gemini & Google Search

This n8n workflow replicates Google's Gemini fullstack LangGraph quickstart, enabling Perplexity-style AI research. It starts with a user query, uses Gemini 2.0 Flash to generate initial search terms, performs web searches via Google Search tool integrated with Gemini, then reflects on results to identify knowledge gaps and create follow-up queries. The process loops iteratively (controlled by parameters like max_research_loops) until a comprehensive, well-supported answer with citations is produced. Redis maintains state for counters and results across loops.

Key benefits include time-saving automated deep research, reducing manual searching and synthesis. It's highly accurate due to reflection cycles that fill gaps, mimicking advanced AI agents. Ideal for content creators, analysts, marketers needing quick, cited insights on complex topics.

Use cases: Market research, academic summaries, competitive analysis, blog post fact-checking. Customize by swapping Gemini models, adjusting query counts, or integrating other tools. Requires Gemini API credentials and Redis setup for seamless operation in n8n.