the paper aims at proposing a generic software framework based on emergence as main paradigm (within a blend with conventional ones, to attain a lever effect) and on non-algorithmic approaches to treat uncertainty at lower echelons (to suit complex application requirements even in very dynamic environments). This target is split into four specific objectives: a) investigating the relationships between complexity and emergence from the standpoint of modern artificial intelligence; b) showing that structural complexity can be dealt with through simulated emergence (e.g., via stigmergic coordination) and cognitive complexity through emulated emergence (e.g., via self-aware agents); c) investigating the (in)adequacy of logics and prediction methods used to handle uncertainty due to future contingents; d) outlining the path for developing affordable non-algorithmic mechanisms to deal with effectively.