1 comment

[ 0.20 ms ] story [ 6.6 ms ] thread
Hmm any way a similar method could be used to do efficient extraction from other things besides agent traces?

I’m thinking logs, documents, etc.