Invisible Threads
Find the signal in the noise
13 followers
Find the signal in the noise
13 followers
Discover hidden connections between ideas across podcasts, essays, and books. AI-powered insight extraction reveals the invisible threads linking Lenny's Podcast, Paul Graham's Essays, and more.
How do you deal with context dependence? Podcasts are conversational - the same phrase can mean the opposite in different episodes depending on what was discussed before. Have you tried accounting for the mood or direction of the conversation, not just the extracted insights?
Also, the design is gorgeous! so light and effortless that you forget it’s an AI tool and not an aesthetic project
@arinam - context dependence is a much harder problem and this project doesn't account for it. Depending on the source (YT transcript vs an essay or a letter) - I have changed the size of each chunk. This allows for a larger capture of context and then judged on its rarity and then the insight is embedded (based on topic and not its semantics - allows for topic matching that is important for thread creation).
And thanks - the design took almost the same if not higher effort compared to the insight creation.
Cloudthread
Interesting idea - do you input media you consume and get insightful connections or does Invisible Threads suggest new media you might like?
@daniele_packard - the media I add gets turned into insightful connections