The Open Economic Framework (OEF) and Decentralized Ledger Technologies (DLTs) allow AEAs to create value through their interaction with other AEAs. The following diagram illustrates the relation of AEAs to the OEF and DLTs.
Open Economic Framework (OEF)
The 'Open Economic Framework' (OEF) consists of protocols, languages and market mechanisms agents use to search and find each other, communicate with as well as trade with each other. As such the OEF defines the decentralised virtual environment that supplies and supports APIs for autonomous third-party software agents, also known as Autonomous Economic Agents (AEAs).
The OEF is under development. Expect frequent changes. What follows is a description of the current implementation.
At present, the OEF's capabilities are fulfilled by two components:
- a permissionless, public peer to peer (agent to agent) communication network, called the Agent Communication Network;
- a set of agent interaction protocols; and
- a centralized search and discovery system.
The latter will be decentralized over time.
Agent Communication Network (ACN)
The agent communication network is a peer-to-peer communication network for agents. It allows agents, in particular AEAs, to send and receive envelopes between each other.
The implementation builds on the open-source libp2p library. A distributed hash table is used by all participating peers to maintain a mapping between agents' cryptographic addresses and their network addresses.
Agents can receive messages from other agents if they are both connected to the ACN (see here for an example).
Centralized search and discovery
A simple OEF (SOEF) search node allows agents to search and discover each other. In particular, agents can register themselves and their services as well as send search requests.
For two agents to be able to find each other, at least one must register themselves and the other must query the SOEF search node for it. Detailed documentation is provided here.
Ledgers enable the AEAs to complete a transaction, which can involve the transfer of funds to each other or the execution of smart contracts. They optionally ensure the truth and integrity of agent to agent interactions.
Whilst a ledger can, in principle, also be used to store structured data - for instance, training data in a machine learning model - in most use cases the resulting costs and privacy implications do not make this a relevant use of the ledger. Instead, usually only references to the structured data - often in the form of hashes - are stored on the ledger and the actual data is stored off-chain.
The Python version of the AEA Framework currently integrates with three ledgers:
However, the framework makes it straightforward for further ledgers to be added by any developer.
AEAs as second layer technology
The following presentation discusses how AEAs can be seen as second layer technology to ledgers.