The fundamental assumptions of distributed systems is having multiple machines that fail independently, communicate over unreliable networks and have no shared clock has the consequence of needing to solve consensus, byzantine faults, ordering, consistency vs. availability and exactly-once delivery.

However, AI agents don't share these problems in the classical sense. Building agents is about context attention, relevance, and information density inside a single ordered buffer. The distributed part is creating an orchestrator that manages these things. At noetive.io we currently work on the context relevance part with our contextual broker Semantik.