Improved Chat: Knowledge Base + Web 🧠 Curating Knowledge With Recall + AI

The video showcases Recall, a knowledge management tool that combines features from Notion, Obsidian, and NotebookLM into one platform. The main focus is on Recall's new chat functionality that can search three ways: web only, personal knowledge base only, or both combined. The speaker demonstrates how to capture content using Recall's browser extension, which can automatically summarize YouTube videos with timestamps and transcripts. He shows how the system creates connections between notes and visualizes them in a knowledge graph. The core argument is that as AI becomes ubiquitous, personal differentiation comes from curating custom knowledge bases rather than relying on generic LLM outputs. Through practical examples, the speaker demonstrates how asking the same question ('what gives me flow states?') yields generic web results versus personalized answers when filtered through his curated notes. He then shows the hybrid approach, using both his personal knowledge and updated web research to analyze gaps in his understanding of flow states since 2026 research. The system allows users to save chat results as new notes, bulk edit tags, and even connect to external AI agents through APIs. The speaker emphasizes that this approach creates a 'memory layer' separate from any specific AI model, allowing users to maintain their personalized knowledge base as new models emerge.