1/ It’s time to share more about what we’ve been working on: @common_fabric is a social computing lab building a medium for software that revolves around people, not apps. commonfabric.com
Obviously people will want to use AI as a personal assistant. And obviously it will need all of your intimate context to do that job well. And so obviously it will need to be private.
Software has always needed your data in order to do anything for you. So, you have to trust any software you use (and now any AI agent) with your information. But what if software didn’t have to have your data to do useful things for you?
LLMs have a larger “working memory” than humans. We have a working memory set of 7 +/- 2 (though we chunk things and draw on vast stores of background info). LLMs have a million token context window. It’s like having infinite pencil and paper, time, and patience.
Someone I know asked Astra to play Factorio. It wins, but it takes forever and guzzles tokens. If you ask Astra instead to “design a system to win Factorio,” it builds software that runs 1000x faster. Mechanistic software gives you cheap leverage, and LLMs can write mechanistic software on demand.