Sharing ideas well is harder than it looks. Generative AI can produce polished output, but it consistently misses what matters most: genuine insight, depth, and a distinct point of view. That’s why I still write everything by hand. What interests me instead is building tools that sharpen each phase of the process, from how ideas take shape to how they reach people.
AI might be useful here though, not as a passive bookmark, but as something more proactive: connecting captured ideas with broader knowledge, live web context, or other content you’ve already created. The boundary between what you wrote and what was generated has to stay visible, not through separate sections or files, but as annotated fragments that can be edited or commented inline.
A tool like this would need real-time data access, through agentic browser use or focused services like Firecrawl or the xAI API, and the ability to process and generate multimodal content. It could live as an MCP or CLI tool, interacting with a filesystem to gather and enrich notes over time.