Agentic AI multi step reasoning agent

1. Input Interpretation & Plan Generation

• Intent Extraction: The agent receives a complex research prompt, decomposes it, and identifies key entities and required information types.
• Dynamic Planning: The agent creates a research plan, generating specific search queries (e.g., academic, blog, and news queries) to gather comprehensive, diverse perspectives. 

2. Recursive Information Gathering (Multi-Hop)

• Intelligent Browsing: The agent conducts parallel searches, visiting blogs, technical articles, and academic papers simultaneously.
• Recursive Reasoning: It reads initial results, learns from them, and reformulates its plan to dig deeper into promising subtopics.

Content Scraping & Extraction: Using tools to extract raw text and ignore ads/navigation, specifically focusing on PDF parsing for academic content.