Consensus Mechanism

DeAgent ensures consensus through

• Entropy-based decision selection:Selecting the lowest-entropy output from candidate AI responses to minimize ambiguity.
• Zero-knowledge proofs (zk):Validating model calls (e.g., via zkTLS for API interactions) and non-model logic to guarantee correctness
• Hybrid active/PoW schemes: For open-source models, using a network of nodes to compute results and selecting the lowest-entropy outcome.

Identity Uniqueness

Each DeAgent inherits the distributed system’s uniqueness properties (e.g., immutability of on-chain data), ensuring:

• Only one valid interaction outcome exists at any time.
• Conflicting actions (e.g., simultaneous contradictory decisions) are prevented, akin to a blockchain's linear chain design.

Governance Capabilities

DeAgent enables governance through

• Decision plugins:Allowing agents to propose, simulate, and execute decisions (e.g., on-chain transactions) with approval/rejection mechanisms.
• Memory and tool integration:Leveraging memory (hot/long-term) and tools (e.g., distributed data queries, internet access) to support informed, context-aware governance.
• DAO frameworks:Supporting multi-agent governance, where agents can vote on proposals and coordinate actions within decentralized systems.