Comment by pkkkzip

what are the implicaitons and actual real world application of this? better agents? more accurate, debuggable LLM answers?