Harrison Chase says defining good agent behavior will decide winners in domains like biology

LangChain's Harrison Chase says whoever can define what good looks like for an AI agent in a given domain will be best positioned to win that domain.

He names biology as an example and says he is bullish on Benchling, citing the company's domain focus. In a quoted post, @nlarusstone says frontier models still struggle with basic molecular cloning tasks. The report contains only these statements from Chase and the quoted post, with no data on Benchling's products or results.