> I think solving memory is solving agi.

But we already have AGI