@yoonholeee i don't understand, isn't anthropic already training their models to a carefully selected harness design. Isn't this benchmarks overfitted then? What is the use then? Fitting them to a specific type of tasks? Anyone using this?
@GergelyOrosz What about SaaS that try to replace their UI interactions with an embedded agent? For instance in healthcare, itโs not realistic to ask every employee to have a claude + MCP hence itโs easier to just offer your agent
@BetterCallMedhi Je trouve la comparison anthropic psg un peu maladroite, anthropic change le monde dans lequel on vit, le psg est juste le meilleur club football actual
Met Ana and went to watch Project Hail with her. She didnโt watch the movie, she lived through it. It was fascinating to see her being shocked with a high pitch sound, laugh by turning around and coming closer to me.