Ever left a meeting feeling like the summary missed the mark? That's because traditional meeting summarization tools often rely solely on transcripts, overlooking valuable context from presentations, notes, and other materials. New research explores how Large Language Models (LLMs) can create richer, more insightful summaries by incorporating these additional resources. The study proposes a three-stage LLM-based approach: First, identify sections of the transcript that lack context. Second, use retrieval augmented generation (RAG) to pull relevant information from supplementary materials to fill these gaps. Finally, generate a comprehensive summary from the enriched transcript. The results are impressive, with summaries showing nearly 10% improvement in relevance and greater content richness. The researchers also introduce a personalization protocol where the LLM analyzes participant characteristics to tailor summaries to individual needs, further boosting informativeness by about 10%. Tested across various LLMs, including smaller models suitable for edge devices, the pipeline demonstrates promising results for real-world applications. While more work is needed to refine aspects like multi-agent retrieval and persona-content matching, this research paves the way for AI-powered meeting summarization that truly captures the essence of complex discussions.