If, in our heart, we still cling to anything - anger, anxiety, possessions - we cannot be free

– Thich Naht Hanh

went to confluent ai and built re:sync, a news aggregator from reddit, arxiv, hugging face, and github repos to help companies or individuals keep up with the AI hype. it'll fetch you recommended sources that are most relevant to your current project. honestly integrating AI news by swyx would be enough, it already summarizes top ai discord, reddit, twitter.

spent 5 bucks to "claude" the entire front end. faced ./zlib_bindings issues and decided to just hard code stuff till the very last minute. demo'ed at the event, the audience was so dead. they laughed and got excited about a buggy VR demo instead, which won first prize in the end.

didn't win but some founders came to talk to us about joining their company. we need connections now more than ever now.

currently working on building knowledge graphs with gemini for the Google - Gemini Long Context kaggle comp. does it make sense to fit a million arxiv paper into gemini in batches? and have it merge knowledge graphs? what will be the killer use case of the context for knowledge graphs? can you generate better knowledge graphs with more in the context? maybe cross-domain ideas? i need to talk to more people about this project, what would be the problem to solve here? goal is to create a kaggle notebook and a youtube video. going to be fun.