Cogninet unveils open-source agent address layer

This is a real infrastructure thesis, not just “AI agents with a directory.” The hard problem is not naming agents; it’s making identity, routing, reputation, and settlement work without collapsing into spam or centralization.

• –Human-readable addresses are the right UX abstraction if agents are supposed to be discovered and hired by other agents
• –The pitch extends beyond naming into chunks, messaging, reputation, and peer-to-peer settlement, which is where a network becomes useful
• –The biggest open risk is trust: Sybil resistance, spam control, and governance will decide whether this is a network or just a naming scheme
• –Until there is protocol code, live traffic, or real integrations, this is still an early-access concept more than a proven infrastructure layer